correlation_analysis.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tqdm import tqdm \n",
    "from xml.etree import ElementTree as ET\n",
    "from datetime import datetime\n",
    "import re\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics.cluster import entropy, mutual_info_score, normalized_mutual_info_score, adjusted_mutual_info_score\n",
    "from sklearn.feature_selection import mutual_info_regression\n",
    "from sklearn.feature_selection import VarianceThreshold\n",
    "MI = lambda x, y: mutual_info_score(x, y)\n",
    "NMI = lambda x, y: mutual_info_regression(x, y)#normalized_mutual_info_score(x, y, average_method='arithmetic')\n",
    "AMI = lambda x, y: adjusted_mutual_info_score(x, y, average_method='arithmetic')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "task_name = 'trial-duration-prediction/trial-duration-prediction_MI.csv'\n",
    "target = pd.read_csv(task_name, index_col=0)\n",
    "target = target[target['Not NA'] > target['Not NA']['required_header/download_date'] * 0.2]\n",
    "target = target[target['var'].fillna(-1) != 0]\n",
    "target\n",
    "target.to_csv(task_name.replace('.csv', '_filtered.csv'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "All_data = []\n",
    "for i in range(10):\n",
    "    All_data.append(pd.read_csv(f'../../preprocess_data/All_data_part{i}.csv', index_col=0, low_memory=False))\n",
    "    # print(len(All_data[i].columns))\n",
    "    \n",
    "All_Ori = pd.concat(All_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_info = pd.read_csv('feature_info.csv', index_col=0)\n",
    "features_category = feature_info[feature_info['len(key)'] < 15].index\n",
    "features_continuous = ['eligibility/minimum_age', 'eligibility/maximum_age', 'number_of_arms']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "from numpy import nan\n",
    "def refine_age(value):\n",
    "    if pd.isnull(value):\n",
    "        return None\n",
    "    elif 'Year' in value:\n",
    "        return int(value.split(' ')[0])\n",
    "    elif 'Month' in value:\n",
    "        return int(value.split(' ')[0]) / 12\n",
    "    elif 'Week' in value:\n",
    "        return int(value.split(' ')[0]) / 52\n",
    "    elif 'Day' in value:\n",
    "        return int(value.split(' ')[0]) / 365\n",
    "    else:\n",
    "        return None\n",
    "\n",
    "All_Ori['eligibility/minimum_age'] = All_Ori['eligibility/minimum_age'].apply(refine_age)\n",
    "All_Ori['eligibility/maximum_age'] = All_Ori['eligibility/maximum_age'].apply(refine_age)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## adverse-event-rate-prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "required_header/download_date  非NA数量: 31306  Var:  0.0  MI:  0.003770619271152409\n",
      "required_header/link_text  非NA数量: 31306  Var:  0.0  MI:  0\n",
      "sponsors/lead_sponsor/agency_class  非NA数量: 31306  Var:  1.0093108922010927  MI:  0.08698691984091722\n",
      "overall_status  非NA数量: 31306  Var:  1.2854321348110327  MI:  0.02592061681115032\n",
      "phase  非NA数量: 20589  Var:  2.4542424500726763  MI:  0.08321975746523114\n",
      "study_type  非NA数量: 31306  Var:  0.04693039681579317  MI:  0.0048368019356219705\n",
      "has_expanded_access  非NA数量: 30882  Var:  0.008921260382614754  MI:  0.0052427723820343\n",
      "study_design_info/intervention_model  非NA数量: 30008  Var:  1.2978838972941327  MI:  0.07773698165878029\n",
      "study_design_info/primary_purpose  非NA数量: 30061  Var:  4.632484584516253  MI:  0.06539074543025114\n",
      "eligibility/gender  非NA数量: 31306  Var:  0.2087674959304742  MI:  0.0044778127862872275\n",
      "eligibility/healthy_volunteers  非NA数量: 31261  Var:  0.16618690719906895  MI:  0.07911606445455899\n",
      "study_first_posted{type}  非NA数量: 31306  Var:  0.23690124508189425  MI:  0.02116262353670839\n",
      "last_update_posted{type}  非NA数量: 31306  Var:  0.0011167823070034701  MI:  0.004428613084292987\n",
      "study_design_info/observational_model  非NA数量: 1238  Var:  1.9472073375708339  MI:  0.05277852000133221\n",
      "study_design_info/time_perspective  非NA数量: 1238  Var:  0.5083119695234314  MI:  0.06888182884075755\n",
      "oversight_info/has_dmc  非NA数量: 28344  Var:  0.24825791616904047  MI:  0.03453474646914323\n",
      "completion_date{type}  非NA数量: 31244  Var:  0.029932947458786656  MI:  0.010195778082029872\n",
      "primary_completion_date{type}  非NA数量: 31306  Var:  0.0006384674300133954  MI:  0\n",
      "enrollment{type}  非NA数量: 31306  Var:  9.582215349537806e-05  MI:  0.0006542514997476268\n",
      "biospec_retention  非NA数量: 204  Var:  0.3391046073601855  MI:  0\n",
      "eligibility/sampling_method  非NA数量: 1238  Var:  0.18825641273444146  MI:  0\n",
      "responsible_party/responsible_party_type  非NA数量: 31287  Var:  0.25250452655387606  MI:  0.04591965404174436\n",
      "last_known_status  非NA数量: 68  Var:  0.0  MI:  0\n",
      "study_design_info/allocation  非NA数量: 21822  Var:  0.12020482590179718  MI:  0.039100321576947206\n",
      "start_date{type}  非NA数量: 23910  Var:  0.0  MI:  0\n",
      "results_first_posted{type}  非NA数量: 31306  Var:  0.018860367881194533  MI:  0.0002888874129651775\n",
      "clinical_results/baseline/analyzed_list/analyzed/scope  非NA数量: 31306  Var:  0.00820499114556528  MI:  0.000936719355429716\n",
      "clinical_results/reported_events/serious_events/default_assessment  非NA数量: 16779  Var:  0.1841671530120153  MI:  0.015295441975538004\n",
      "clinical_results/reported_events/other_events/default_assessment  非NA数量: 21943  Var:  0.18957059687957034  MI:  0.013166912521107133\n",
      "clinical_results/certain_agreements/pi_employee  非NA数量: 31306  Var:  0.20837011393264154  MI:  0.03802348315871473\n",
      "patient_data/sharing_ipd  非NA数量: 18003  Var:  0.7839436307702744  MI:  0.04013465337549871\n",
      "oversight_info/is_fda_regulated_drug  非NA数量: 17967  Var:  0.24137924240379324  MI:  0.07213317412133646\n",
      "oversight_info/is_fda_regulated_device  非NA数量: 17953  Var:  0.14449977972787534  MI:  0.019618290050650078\n",
      "oversight_info/is_us_export  非NA数量: 5358  Var:  0.23491640503719394  MI:  0.020936863904410963\n",
      "disposition_first_posted{type}  非NA数量: 3045  Var:  0.19638363660294872  MI:  0\n",
      "eligibility/gender_based  非NA数量: 694  Var:  0.0  MI:  0.03684329583541812\n",
      "expanded_access_info/expanded_access_type_intermediate  非NA数量: 0  Var:  nan  MI:  N/A\n",
      "expanded_access_info/expanded_access_type_treatment  非NA数量: 0  Var:  nan  MI:  N/A\n",
      "oversight_info/is_unapproved_device  非NA数量: 686  Var:  0.0  MI:  0.005153524455911995\n",
      "expanded_access_info/expanded_access_type_individual  非NA数量: 0  Var:  nan  MI:  N/A\n",
      "oversight_info/is_ppsd  非NA数量: 2  Var:  0.0  MI:  N/A\n",
      "eligibility/minimum_age  非NA数量: 30146  Var:  108.49367558905443  MI:  0.029924567279308256\n",
      "eligibility/maximum_age  非NA数量: 15129  Var:  766.7921322984097  MI:  0.06202449933894716\n",
      "number_of_arms  非NA数量: 30022  Var:  2.8206703212620923  MI:  0.055431720901276904\n",
      "required_header/url 非NA数量: 31306\n",
      "id_info/org_study_id 非NA数量: 31306\n",
      "id_info/secondary_id 非NA数量: 12042\n",
      "id_info/nct_id 非NA数量: 31306\n",
      "brief_title 非NA数量: 31306\n",
      "sponsors/lead_sponsor/agency 非NA数量: 31306\n",
      "source 非NA数量: 31306\n",
      "brief_summary/textblock 非NA数量: 31306\n",
      "detailed_description/textblock 非NA数量: 21875\n",
      "study_design_info/masking 非NA数量: 30030\n",
      "condition 非NA数量: 31306\n",
      "intervention/intervention_type 非NA数量: 30894\n",
      "intervention/intervention_name 非NA数量: 30894\n",
      "eligibility/criteria/textblock 非NA数量: 31305\n",
      "location/facility/name 非NA数量: 29374\n",
      "location/facility/address/city 非NA数量: 30157\n",
      "location/facility/address/state 非NA数量: 27011\n",
      "location/facility/address/country 非NA数量: 30157\n",
      "location_countries/country 非NA数量: 30157\n",
      "verification_date 非NA数量: 31306\n",
      "study_first_submitted 非NA数量: 31306\n",
      "study_first_submitted_qc 非NA数量: 31306\n",
      "study_first_posted 非NA数量: 31306\n",
      "last_update_submitted 非NA数量: 31306\n",
      "last_update_submitted_qc 非NA数量: 31306\n",
      "last_update_posted 非NA数量: 31306\n",
      "condition_browse/mesh_term 非NA数量: 27199\n",
      "intervention_browse/mesh_term 非NA数量: 16056\n",
      "icdcode 非NA数量: 31306\n",
      "sponsors/collaborator/agency 非NA数量: 12052\n",
      "sponsors/collaborator/agency_class 非NA数量: 12052\n",
      "location/facility/address/zip 非NA数量: 28280\n",
      "keyword 非NA数量: 19196\n",
      "official_title 非NA数量: 31136\n",
      "why_stopped 非NA数量: 4666\n",
      "start_date 非NA数量: 31305\n",
      "completion_date 非NA数量: 31246\n",
      "primary_completion_date 非NA数量: 31306\n",
      "primary_outcome/measure 非NA数量: 31306\n",
      "primary_outcome/time_frame 非NA数量: 31305\n",
      "secondary_outcome/measure 非NA数量: 25291\n",
      "secondary_outcome/time_frame 非NA数量: 25291\n",
      "number_of_groups 非NA数量: 966\n",
      "enrollment 非NA数量: 31306\n",
      "arm_group/arm_group_label 非NA数量: 30988\n",
      "arm_group/description 非NA数量: 29231\n",
      "intervention/description 非NA数量: 28599\n",
      "intervention/arm_group_label 非NA数量: 30720\n",
      "intervention/other_name 非NA数量: 16023\n",
      "biospec_descr/textblock 非NA数量: 204\n",
      "eligibility/study_pop/textblock 非NA数量: 1238\n",
      "overall_official/last_name 非NA数量: 27294\n",
      "overall_official/role 非NA数量: 27290\n",
      "overall_official/affiliation 非NA数量: 27259\n",
      "overall_contact/last_name 非NA数量: 3\n",
      "overall_contact/phone 非NA数量: 3\n",
      "overall_contact/email 非NA数量: 3\n",
      "location/status 非NA数量: 3\n",
      "arm_group/arm_group_type 非NA数量: 30022\n",
      "link/url 非NA数量: 5649\n",
      "link/description 非NA数量: 5649\n",
      "results_reference/citation 非NA数量: 2868\n",
      "results_reference/PMID 非NA数量: 2742\n",
      "reference/citation 非NA数量: 4786\n",
      "reference/PMID 非NA数量: 4681\n",
      "study_design_info/masking_description 非NA数量: 2435\n",
      "primary_outcome/description 非NA数量: 30333\n",
      "secondary_outcome/description 非NA数量: 24671\n",
      "responsible_party/investigator_affiliation 非NA数量: 9185\n",
      "responsible_party/investigator_full_name 非NA数量: 9185\n",
      "responsible_party/investigator_title 非NA数量: 9185\n",
      "acronym 非NA数量: 8063\n",
      "results_first_submitted 非NA数量: 31306\n",
      "results_first_submitted_qc 非NA数量: 31306\n",
      "results_first_posted 非NA数量: 31306\n",
      "provided_document_section/provided_document/document_type 非NA数量: 23174\n",
      "provided_document_section/provided_document/document_has_protocol 非NA数量: 23174\n",
      "provided_document_section/provided_document/document_has_icf 非NA数量: 23174\n",
      "provided_document_section/provided_document/document_has_sap 非NA数量: 23174\n",
      "provided_document_section/provided_document/document_date 非NA数量: 23174\n",
      "provided_document_section/provided_document/document_url 非NA数量: 23174\n",
      "clinical_results/participant_flow/group_list/group{group_id} 非NA数量: 31306\n",
      "clinical_results/participant_flow/group_list/group/title 非NA数量: 31306\n",
      "clinical_results/participant_flow/group_list/group/description 非NA数量: 31280\n",
      "clinical_results/participant_flow/period_list/period/title 非NA数量: 31306\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/title 非NA数量: 31306\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/participants_list/participants{group_id} 非NA数量: 31306\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/participants_list/participants{count} 非NA数量: 31306\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/participants_list/participants 非NA数量: 2892\n",
      "clinical_results/baseline/population 非NA数量: 11357\n",
      "clinical_results/baseline/group_list/group{group_id} 非NA数量: 31306\n",
      "clinical_results/baseline/group_list/group/title 非NA数量: 31306\n",
      "clinical_results/baseline/group_list/group/description 非NA数量: 31269\n",
      "clinical_results/baseline/analyzed_list/analyzed/units 非NA数量: 31306\n",
      "clinical_results/baseline/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 31306\n",
      "clinical_results/baseline/analyzed_list/analyzed/count_list/count{value} 非NA数量: 31306\n",
      "clinical_results/baseline/measure_list/measure/title 非NA数量: 31306\n",
      "clinical_results/baseline/measure_list/measure/units 非NA数量: 31292\n",
      "clinical_results/baseline/measure_list/measure/param 非NA数量: 31250\n",
      "clinical_results/baseline/measure_list/measure/dispersion 非NA数量: 26669\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{group_id} 非NA数量: 31249\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{value} 非NA数量: 31249\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{spread} 非NA数量: 21092\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/title 非NA数量: 31122\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/title 非NA数量: 22588\n",
      "clinical_results/outcome_list/outcome/type 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/title 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/description 非NA数量: 30802\n",
      "clinical_results/outcome_list/outcome/time_frame 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/population 非NA数量: 24041\n",
      "clinical_results/outcome_list/outcome/group_list/group{group_id} 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/group_list/group/title 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/group_list/group/description 非NA数量: 31272\n",
      "clinical_results/outcome_list/outcome/measure/title 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/measure/description 非NA数量: 30796\n",
      "clinical_results/outcome_list/outcome/measure/population 非NA数量: 24041\n",
      "clinical_results/outcome_list/outcome/measure/units 非NA数量: 30507\n",
      "clinical_results/outcome_list/outcome/measure/param 非NA数量: 30507\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/units 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/scope 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/count_list/count{value} 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/title 非NA数量: 16048\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{group_id} 非NA数量: 30522\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{value} 非NA数量: 30517\n",
      "clinical_results/reported_events/time_frame 非NA数量: 29618\n",
      "clinical_results/reported_events/desc 非NA数量: 14403\n",
      "clinical_results/reported_events/group_list/group{group_id} 非NA数量: 31306\n",
      "clinical_results/reported_events/group_list/group/title 非NA数量: 31306\n",
      "clinical_results/reported_events/group_list/group/description 非NA数量: 31232\n",
      "clinical_results/reported_events/serious_events/category_list/category/title 非NA数量: 31306\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/sub_title 非NA数量: 31306\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{group_id} 非NA数量: 31306\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{subjects_affected} 非NA数量: 31306\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{subjects_at_risk} 非NA数量: 31306\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/description 非NA数量: 2950\n",
      "clinical_results/reported_events/other_events/frequency_threshold 非NA数量: 31306\n",
      "clinical_results/reported_events/other_events/category_list/category/title 非NA数量: 31306\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/sub_title 非NA数量: 31306\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{group_id} 非NA数量: 31306\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{subjects_affected} 非NA数量: 31306\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{subjects_at_risk} 非NA数量: 31306\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/description 非NA数量: 3647\n",
      "clinical_results/certain_agreements/restrictive_agreement 非NA数量: 30796\n",
      "clinical_results/point_of_contact/name_or_title 非NA数量: 31306\n",
      "clinical_results/point_of_contact/organization 非NA数量: 31306\n",
      "clinical_results/point_of_contact/phone 非NA数量: 30792\n",
      "clinical_results/point_of_contact/email 非NA数量: 30720\n",
      "clinical_results/participant_flow/recruitment_details 非NA数量: 11754\n",
      "clinical_results/reported_events/serious_events/default_vocab 非NA数量: 11596\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{events} 非NA数量: 10267\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/sub_title{vocab} 非NA数量: 248\n",
      "clinical_results/reported_events/other_events/default_vocab 非NA数量: 13886\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{events} 非NA数量: 13330\n",
      "clinical_results/participant_flow/pre_assignment_details 非NA数量: 11794\n",
      "clinical_results/participant_flow/period_list/period/drop_withdraw_reason_list/drop_withdraw_reason/title 非NA数量: 18725\n",
      "clinical_results/participant_flow/period_list/period/drop_withdraw_reason_list/drop_withdraw_reason/participants_list/participants{group_id} 非NA数量: 18725\n",
      "clinical_results/participant_flow/period_list/period/drop_withdraw_reason_list/drop_withdraw_reason/participants_list/participants{count} 非NA数量: 18725\n",
      "id_info/nct_alias 非NA数量: 377\n",
      "responsible_party/name_title 非NA数量: 7\n",
      "responsible_party/organization 非NA数量: 8\n",
      "location/contact/last_name 非NA数量: 1\n",
      "location/contact/phone 非NA数量: 1\n",
      "location/contact/email 非NA数量: 1\n",
      "removed_countries/country 非NA数量: 2746\n",
      "clinical_results/outcome_list/outcome/measure/dispersion 非NA数量: 24981\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{spread} 非NA数量: 17622\n",
      "clinical_results/limitations_and_caveats 非NA数量: 6190\n",
      "other_outcome/measure 非NA数量: 2966\n",
      "other_outcome/time_frame 非NA数量: 2966\n",
      "pending_results/submitted 非NA数量: 0\n",
      "pending_results/returned 非NA数量: 0\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/group_id_list/group_id 非NA数量: 10562\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/groups_desc 非NA数量: 5354\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/non_inferiority_type 非NA数量: 10562\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/p_value 非NA数量: 9038\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/p_value_desc 非NA数量: 3244\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/method 非NA数量: 9038\n",
      "clinical_results/baseline/measure_list/measure/description 非NA数量: 7470\n",
      "study_design_info/intervention_model_description 非NA数量: 4234\n",
      "study_docs/study_doc/doc_id 非NA数量: 263\n",
      "study_docs/study_doc/doc_type 非NA数量: 342\n",
      "study_docs/study_doc/doc_url 非NA数量: 342\n",
      "study_docs/study_doc/doc_comment 非NA数量: 284\n",
      "patient_data/ipd_description 非NA数量: 6361\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{lower_limit} 非NA数量: 12744\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{upper_limit} 非NA数量: 12697\n",
      "other_outcome/description 非NA数量: 2830\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{lower_limit} 非NA数量: 6645\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{upper_limit} 非NA数量: 6626\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/param_type 非NA数量: 7100\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/param_value 非NA数量: 7100\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_percent 非NA数量: 6415\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_n_sides 非NA数量: 7042\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_lower_limit 非NA数量: 6265\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_upper_limit 非NA数量: 6276\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/method_desc 非NA数量: 2414\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/assessment 非NA数量: 315\n",
      "overall_contact_backup/last_name 非NA数量: 2\n",
      "overall_contact_backup/phone 非NA数量: 2\n",
      "overall_contact_backup/email 非NA数量: 1\n",
      "location/contact/phone_ext 非NA数量: 1\n",
      "patient_data/ipd_info_type 非NA数量: 2625\n",
      "patient_data/ipd_time_frame 非NA数量: 2497\n",
      "patient_data/ipd_access_criteria 非NA数量: 2611\n",
      "clinical_results/outcome_list/outcome/measure/units_analyzed 非NA数量: 1450\n",
      "clinical_results/baseline/measure_list/measure/population 非NA数量: 6162\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/units 非NA数量: 5583\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/scope 非NA数量: 5583\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 5583\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/count_list/count{value} 非NA数量: 5583\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/title 非NA数量: 3204\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/assessment 非NA数量: 641\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement 非NA数量: 3479\n",
      "disposition_first_submitted 非NA数量: 3045\n",
      "disposition_first_submitted_qc 非NA数量: 3045\n",
      "disposition_first_posted 非NA数量: 3045\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/estimate_desc 非NA数量: 2629\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/sub_title{vocab} 非NA数量: 282\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/non_inferiority_desc 非NA数量: 3353\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/other_analysis_desc 非NA数量: 569\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/units 非NA数量: 7140\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/scope 非NA数量: 7140\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 7140\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/count_list/count{value} 非NA数量: 7140\n",
      "clinical_results/baseline/measure_list/measure/units_analyzed 非NA数量: 260\n",
      "patient_data/ipd_url 非NA数量: 2584\n",
      "location/contact_backup/last_name 非NA数量: 0\n",
      "location/contact_backup/phone 非NA数量: 0\n",
      "location/contact_backup/email 非NA数量: 0\n",
      "eligibility/gender_description 非NA数量: 436\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/dispersion_type 非NA数量: 1690\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/dispersion_value 非NA数量: 1690\n",
      "location/investigator/last_name 非NA数量: 1\n",
      "location/investigator/role 非NA数量: 1\n",
      "target_duration 非NA数量: 96\n",
      "pending_results/submission_canceled 非NA数量: 0\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement 非NA数量: 317\n",
      "clinical_results/outcome_list/outcome/posting_date 非NA数量: 436\n",
      "overall_contact/phone_ext 非NA数量: 0\n",
      "location/contact_backup/phone_ext 非NA数量: 0\n",
      "overall_contact_backup/phone_ext 非NA数量: 0\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_upper_limit_na_comment 非NA数量: 24\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/units 非NA数量: 84\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/scope 非NA数量: 84\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 84\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/count_list/count{value} 非NA数量: 84\n"
     ]
    }
   ],
   "source": [
    "task_name = 'adverse-event-rate-prediction'\n",
    "target = pd.read_csv(f'{task_name}/output.csv', index_col=0)\n",
    "All = All_Ori.loc[target.index]\n",
    "target = target['serious_adverse_rate']\n",
    "MI_VAR = []\n",
    "for feature_name in features_category:\n",
    "    print(feature_name, end='  ')\n",
    "    notna = All[feature_name].notna().sum()\n",
    "    print('非NA数量:', notna, end='  ')\n",
    "    var = All[All[feature_name].notna()][feature_name].astype('category').cat.codes.var()\n",
    "    print('Var: ', var, end='  ')\n",
    "    if notna < 5:\n",
    "        mi = 'N/A'\n",
    "        print('MI: ', mi)\n",
    "    else:\n",
    "        mi = NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes.to_numpy().reshape(-1, 1), target.loc[All[All[feature_name].notna()].index])[0]\n",
    "        print('MI: ', mi)\n",
    "\n",
    "    MI_VAR.append((feature_name, notna, var, mi, 'Category'))\n",
    "\n",
    "for feature_name in features_continuous:\n",
    "    print(feature_name, end='  ')\n",
    "    notna = All[feature_name].notna().sum()\n",
    "    print('非NA数量:', notna, end='  ')\n",
    "    var = All[All[feature_name].notna()][feature_name].var()\n",
    "    print('Var: ', var, end='  ')\n",
    "    if notna < 5:\n",
    "        mi = 'N/A'\n",
    "        print('MI: ', mi)\n",
    "    else:\n",
    "        mi = NMI(All[All[feature_name].notna()][feature_name].to_numpy().reshape(-1, 1), target.loc[All[All[feature_name].notna()].index])[0]\n",
    "        print('MI: ', mi)\n",
    "\n",
    "    MI_VAR.append((feature_name, notna, var, mi, 'Continues'))\n",
    "\n",
    "df = pd.DataFrame(MI_VAR, columns=['feature_name', 'Not NA', 'var', 'MI', 'Category/Continues']).set_index('feature_name')\n",
    "\n",
    "for feature_name in feature_info.index:\n",
    "    if feature_name not in df.index:\n",
    "        df.loc[feature_name] = (All[feature_name].notna().sum(), None, None, None)\n",
    "        print(feature_name, '非NA数量:', All[feature_name].notna().sum())\n",
    "        \n",
    "df.to_csv(f'{task_name}/serious_adverse_rate_MI.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "required_header/download_date  非NA数量: 31306  Var:  0.0  MI:  0\n",
      "required_header/link_text  非NA数量: 31306  Var:  0.0  MI:  0.005321291583228138\n",
      "sponsors/lead_sponsor/agency_class  非NA数量: 31306  Var:  1.0093108922010927  MI:  0.02941216781870004\n",
      "overall_status  非NA数量: 31306  Var:  1.2854321348110327  MI:  0.021852490260250157\n",
      "phase  非NA数量: 20589  Var:  2.4542424500726763  MI:  0.05614331097131231\n",
      "study_type  非NA数量: 31306  Var:  0.04693039681579317  MI:  0.0025488686689589812\n",
      "has_expanded_access  非NA数量: 30882  Var:  0.008921260382614754  MI:  0.0063649290812808346\n",
      "study_design_info/intervention_model  非NA数量: 30008  Var:  1.2978838972941327  MI:  0.05278978263340406\n",
      "study_design_info/primary_purpose  非NA数量: 30061  Var:  4.632484584516253  MI:  0.03537093044256512\n",
      "eligibility/gender  非NA数量: 31306  Var:  0.2087674959304742  MI:  0.006102674711614853\n",
      "eligibility/healthy_volunteers  非NA数量: 31261  Var:  0.16618690719906895  MI:  0.05198940632596738\n",
      "study_first_posted{type}  非NA数量: 31306  Var:  0.23690124508189425  MI:  0.016484494507941427\n",
      "last_update_posted{type}  非NA数量: 31306  Var:  0.0011167823070034701  MI:  0\n",
      "study_design_info/observational_model  非NA数量: 1238  Var:  1.9472073375708339  MI:  0.004969430808592357\n",
      "study_design_info/time_perspective  非NA数量: 1238  Var:  0.5083119695234314  MI:  0.0263721797517813\n",
      "oversight_info/has_dmc  非NA数量: 28344  Var:  0.24825791616904047  MI:  0.026136645046537232\n",
      "completion_date{type}  非NA数量: 31244  Var:  0.029932947458786656  MI:  0.014345344467309395\n",
      "primary_completion_date{type}  非NA数量: 31306  Var:  0.0006384674300133954  MI:  0.0014497760507321544\n",
      "enrollment{type}  非NA数量: 31306  Var:  9.582215349537806e-05  MI:  0.005317330419568833\n",
      "biospec_retention  非NA数量: 204  Var:  0.3391046073601855  MI:  0\n",
      "eligibility/sampling_method  非NA数量: 1238  Var:  0.18825641273444146  MI:  0\n",
      "responsible_party/responsible_party_type  非NA数量: 31287  Var:  0.25250452655387606  MI:  0.021134409410264787\n",
      "last_known_status  非NA数量: 68  Var:  0.0  MI:  0.15179453304152424\n",
      "study_design_info/allocation  非NA数量: 21822  Var:  0.12020482590179718  MI:  0.025719138634910266\n",
      "start_date{type}  非NA数量: 23910  Var:  0.0  MI:  0\n",
      "results_first_posted{type}  非NA数量: 31306  Var:  0.018860367881194533  MI:  0.0013321025356454186\n",
      "clinical_results/baseline/analyzed_list/analyzed/scope  非NA数量: 31306  Var:  0.00820499114556528  MI:  0\n",
      "clinical_results/reported_events/serious_events/default_assessment  非NA数量: 16779  Var:  0.1841671530120153  MI:  0.0014183054185208377\n",
      "clinical_results/reported_events/other_events/default_assessment  非NA数量: 21943  Var:  0.18957059687957034  MI:  0.004805358406338556\n",
      "clinical_results/certain_agreements/pi_employee  非NA数量: 31306  Var:  0.20837011393264154  MI:  0.015539657025962939\n",
      "patient_data/sharing_ipd  非NA数量: 18003  Var:  0.7839436307702744  MI:  0.011812210064937823\n",
      "oversight_info/is_fda_regulated_drug  非NA数量: 17967  Var:  0.24137924240379324  MI:  0.03640705012931722\n",
      "oversight_info/is_fda_regulated_device  非NA数量: 17953  Var:  0.14449977972787534  MI:  0.012512745532895941\n",
      "oversight_info/is_us_export  非NA数量: 5358  Var:  0.23491640503719394  MI:  0.00013003059774518988\n",
      "disposition_first_posted{type}  非NA数量: 3045  Var:  0.19638363660294872  MI:  0.005841680057981691\n",
      "eligibility/gender_based  非NA数量: 694  Var:  0.0  MI:  0\n",
      "expanded_access_info/expanded_access_type_intermediate  非NA数量: 0  Var:  nan  MI:  N/A\n",
      "expanded_access_info/expanded_access_type_treatment  非NA数量: 0  Var:  nan  MI:  N/A\n",
      "oversight_info/is_unapproved_device  非NA数量: 686  Var:  0.0  MI:  0\n",
      "expanded_access_info/expanded_access_type_individual  非NA数量: 0  Var:  nan  MI:  N/A\n",
      "oversight_info/is_ppsd  非NA数量: 2  Var:  0.0  MI:  N/A\n",
      "eligibility/minimum_age  非NA数量:  30146  Var:  108.49367558905443  MI:  0.034389123938637134\n",
      "eligibility/maximum_age  非NA数量:  15129  Var:  766.7921322984097  MI:  0.0812245982852211\n",
      "number_of_arms  非NA数量:  30022  Var:  2.8206703212620923  MI:  0.032559096712374114\n",
      "required_header/url 非NA数量: 31306\n",
      "id_info/org_study_id 非NA数量: 31306\n",
      "id_info/secondary_id 非NA数量: 12042\n",
      "id_info/nct_id 非NA数量: 31306\n",
      "brief_title 非NA数量: 31306\n",
      "sponsors/lead_sponsor/agency 非NA数量: 31306\n",
      "source 非NA数量: 31306\n",
      "brief_summary/textblock 非NA数量: 31306\n",
      "detailed_description/textblock 非NA数量: 21875\n",
      "study_design_info/masking 非NA数量: 30030\n",
      "condition 非NA数量: 31306\n",
      "intervention/intervention_type 非NA数量: 30894\n",
      "intervention/intervention_name 非NA数量: 30894\n",
      "eligibility/criteria/textblock 非NA数量: 31305\n",
      "location/facility/name 非NA数量: 29374\n",
      "location/facility/address/city 非NA数量: 30157\n",
      "location/facility/address/state 非NA数量: 27011\n",
      "location/facility/address/country 非NA数量: 30157\n",
      "location_countries/country 非NA数量: 30157\n",
      "verification_date 非NA数量: 31306\n",
      "study_first_submitted 非NA数量: 31306\n",
      "study_first_submitted_qc 非NA数量: 31306\n",
      "study_first_posted 非NA数量: 31306\n",
      "last_update_submitted 非NA数量: 31306\n",
      "last_update_submitted_qc 非NA数量: 31306\n",
      "last_update_posted 非NA数量: 31306\n",
      "condition_browse/mesh_term 非NA数量: 27199\n",
      "intervention_browse/mesh_term 非NA数量: 16056\n",
      "icdcode 非NA数量: 31306\n",
      "sponsors/collaborator/agency 非NA数量: 12052\n",
      "sponsors/collaborator/agency_class 非NA数量: 12052\n",
      "location/facility/address/zip 非NA数量: 28280\n",
      "keyword 非NA数量: 19196\n",
      "official_title 非NA数量: 31136\n",
      "why_stopped 非NA数量: 4666\n",
      "start_date 非NA数量: 31305\n",
      "completion_date 非NA数量: 31246\n",
      "primary_completion_date 非NA数量: 31306\n",
      "primary_outcome/measure 非NA数量: 31306\n",
      "primary_outcome/time_frame 非NA数量: 31305\n",
      "secondary_outcome/measure 非NA数量: 25291\n",
      "secondary_outcome/time_frame 非NA数量: 25291\n",
      "number_of_groups 非NA数量: 966\n",
      "enrollment 非NA数量: 31306\n",
      "arm_group/arm_group_label 非NA数量: 30988\n",
      "arm_group/description 非NA数量: 29231\n",
      "intervention/description 非NA数量: 28599\n",
      "intervention/arm_group_label 非NA数量: 30720\n",
      "intervention/other_name 非NA数量: 16023\n",
      "biospec_descr/textblock 非NA数量: 204\n",
      "eligibility/study_pop/textblock 非NA数量: 1238\n",
      "overall_official/last_name 非NA数量: 27294\n",
      "overall_official/role 非NA数量: 27290\n",
      "overall_official/affiliation 非NA数量: 27259\n",
      "overall_contact/last_name 非NA数量: 3\n",
      "overall_contact/phone 非NA数量: 3\n",
      "overall_contact/email 非NA数量: 3\n",
      "location/status 非NA数量: 3\n",
      "arm_group/arm_group_type 非NA数量: 30022\n",
      "link/url 非NA数量: 5649\n",
      "link/description 非NA数量: 5649\n",
      "results_reference/citation 非NA数量: 2868\n",
      "results_reference/PMID 非NA数量: 2742\n",
      "reference/citation 非NA数量: 4786\n",
      "reference/PMID 非NA数量: 4681\n",
      "study_design_info/masking_description 非NA数量: 2435\n",
      "primary_outcome/description 非NA数量: 30333\n",
      "secondary_outcome/description 非NA数量: 24671\n",
      "responsible_party/investigator_affiliation 非NA数量: 9185\n",
      "responsible_party/investigator_full_name 非NA数量: 9185\n",
      "responsible_party/investigator_title 非NA数量: 9185\n",
      "acronym 非NA数量: 8063\n",
      "results_first_submitted 非NA数量: 31306\n",
      "results_first_submitted_qc 非NA数量: 31306\n",
      "results_first_posted 非NA数量: 31306\n",
      "provided_document_section/provided_document/document_type 非NA数量: 23174\n",
      "provided_document_section/provided_document/document_has_protocol 非NA数量: 23174\n",
      "provided_document_section/provided_document/document_has_icf 非NA数量: 23174\n",
      "provided_document_section/provided_document/document_has_sap 非NA数量: 23174\n",
      "provided_document_section/provided_document/document_date 非NA数量: 23174\n",
      "provided_document_section/provided_document/document_url 非NA数量: 23174\n",
      "clinical_results/participant_flow/group_list/group{group_id} 非NA数量: 31306\n",
      "clinical_results/participant_flow/group_list/group/title 非NA数量: 31306\n",
      "clinical_results/participant_flow/group_list/group/description 非NA数量: 31280\n",
      "clinical_results/participant_flow/period_list/period/title 非NA数量: 31306\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/title 非NA数量: 31306\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/participants_list/participants{group_id} 非NA数量: 31306\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/participants_list/participants{count} 非NA数量: 31306\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/participants_list/participants 非NA数量: 2892\n",
      "clinical_results/baseline/population 非NA数量: 11357\n",
      "clinical_results/baseline/group_list/group{group_id} 非NA数量: 31306\n",
      "clinical_results/baseline/group_list/group/title 非NA数量: 31306\n",
      "clinical_results/baseline/group_list/group/description 非NA数量: 31269\n",
      "clinical_results/baseline/analyzed_list/analyzed/units 非NA数量: 31306\n",
      "clinical_results/baseline/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 31306\n",
      "clinical_results/baseline/analyzed_list/analyzed/count_list/count{value} 非NA数量: 31306\n",
      "clinical_results/baseline/measure_list/measure/title 非NA数量: 31306\n",
      "clinical_results/baseline/measure_list/measure/units 非NA数量: 31292\n",
      "clinical_results/baseline/measure_list/measure/param 非NA数量: 31250\n",
      "clinical_results/baseline/measure_list/measure/dispersion 非NA数量: 26669\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{group_id} 非NA数量: 31249\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{value} 非NA数量: 31249\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{spread} 非NA数量: 21092\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/title 非NA数量: 31122\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/title 非NA数量: 22588\n",
      "clinical_results/outcome_list/outcome/type 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/title 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/description 非NA数量: 30802\n",
      "clinical_results/outcome_list/outcome/time_frame 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/population 非NA数量: 24041\n",
      "clinical_results/outcome_list/outcome/group_list/group{group_id} 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/group_list/group/title 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/group_list/group/description 非NA数量: 31272\n",
      "clinical_results/outcome_list/outcome/measure/title 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/measure/description 非NA数量: 30796\n",
      "clinical_results/outcome_list/outcome/measure/population 非NA数量: 24041\n",
      "clinical_results/outcome_list/outcome/measure/units 非NA数量: 30507\n",
      "clinical_results/outcome_list/outcome/measure/param 非NA数量: 30507\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/units 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/scope 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/count_list/count{value} 非NA数量: 31306\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/title 非NA数量: 16048\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{group_id} 非NA数量: 30522\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{value} 非NA数量: 30517\n",
      "clinical_results/reported_events/time_frame 非NA数量: 29618\n",
      "clinical_results/reported_events/desc 非NA数量: 14403\n",
      "clinical_results/reported_events/group_list/group{group_id} 非NA数量: 31306\n",
      "clinical_results/reported_events/group_list/group/title 非NA数量: 31306\n",
      "clinical_results/reported_events/group_list/group/description 非NA数量: 31232\n",
      "clinical_results/reported_events/serious_events/category_list/category/title 非NA数量: 31306\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/sub_title 非NA数量: 31306\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{group_id} 非NA数量: 31306\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{subjects_affected} 非NA数量: 31306\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{subjects_at_risk} 非NA数量: 31306\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/description 非NA数量: 2950\n",
      "clinical_results/reported_events/other_events/frequency_threshold 非NA数量: 31306\n",
      "clinical_results/reported_events/other_events/category_list/category/title 非NA数量: 31306\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/sub_title 非NA数量: 31306\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{group_id} 非NA数量: 31306\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{subjects_affected} 非NA数量: 31306\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{subjects_at_risk} 非NA数量: 31306\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/description 非NA数量: 3647\n",
      "clinical_results/certain_agreements/restrictive_agreement 非NA数量: 30796\n",
      "clinical_results/point_of_contact/name_or_title 非NA数量: 31306\n",
      "clinical_results/point_of_contact/organization 非NA数量: 31306\n",
      "clinical_results/point_of_contact/phone 非NA数量: 30792\n",
      "clinical_results/point_of_contact/email 非NA数量: 30720\n",
      "clinical_results/participant_flow/recruitment_details 非NA数量: 11754\n",
      "clinical_results/reported_events/serious_events/default_vocab 非NA数量: 11596\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{events} 非NA数量: 10267\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/sub_title{vocab} 非NA数量: 248\n",
      "clinical_results/reported_events/other_events/default_vocab 非NA数量: 13886\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{events} 非NA数量: 13330\n",
      "clinical_results/participant_flow/pre_assignment_details 非NA数量: 11794\n",
      "clinical_results/participant_flow/period_list/period/drop_withdraw_reason_list/drop_withdraw_reason/title 非NA数量: 18725\n",
      "clinical_results/participant_flow/period_list/period/drop_withdraw_reason_list/drop_withdraw_reason/participants_list/participants{group_id} 非NA数量: 18725\n",
      "clinical_results/participant_flow/period_list/period/drop_withdraw_reason_list/drop_withdraw_reason/participants_list/participants{count} 非NA数量: 18725\n",
      "id_info/nct_alias 非NA数量: 377\n",
      "responsible_party/name_title 非NA数量: 7\n",
      "responsible_party/organization 非NA数量: 8\n",
      "location/contact/last_name 非NA数量: 1\n",
      "location/contact/phone 非NA数量: 1\n",
      "location/contact/email 非NA数量: 1\n",
      "removed_countries/country 非NA数量: 2746\n",
      "clinical_results/outcome_list/outcome/measure/dispersion 非NA数量: 24981\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{spread} 非NA数量: 17622\n",
      "clinical_results/limitations_and_caveats 非NA数量: 6190\n",
      "other_outcome/measure 非NA数量: 2966\n",
      "other_outcome/time_frame 非NA数量: 2966\n",
      "pending_results/submitted 非NA数量: 0\n",
      "pending_results/returned 非NA数量: 0\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/group_id_list/group_id 非NA数量: 10562\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/groups_desc 非NA数量: 5354\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/non_inferiority_type 非NA数量: 10562\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/p_value 非NA数量: 9038\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/p_value_desc 非NA数量: 3244\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/method 非NA数量: 9038\n",
      "clinical_results/baseline/measure_list/measure/description 非NA数量: 7470\n",
      "study_design_info/intervention_model_description 非NA数量: 4234\n",
      "study_docs/study_doc/doc_id 非NA数量: 263\n",
      "study_docs/study_doc/doc_type 非NA数量: 342\n",
      "study_docs/study_doc/doc_url 非NA数量: 342\n",
      "study_docs/study_doc/doc_comment 非NA数量: 284\n",
      "patient_data/ipd_description 非NA数量: 6361\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{lower_limit} 非NA数量: 12744\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{upper_limit} 非NA数量: 12697\n",
      "other_outcome/description 非NA数量: 2830\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{lower_limit} 非NA数量: 6645\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{upper_limit} 非NA数量: 6626\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/param_type 非NA数量: 7100\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/param_value 非NA数量: 7100\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_percent 非NA数量: 6415\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_n_sides 非NA数量: 7042\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_lower_limit 非NA数量: 6265\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_upper_limit 非NA数量: 6276\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/method_desc 非NA数量: 2414\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/assessment 非NA数量: 315\n",
      "overall_contact_backup/last_name 非NA数量: 2\n",
      "overall_contact_backup/phone 非NA数量: 2\n",
      "overall_contact_backup/email 非NA数量: 1\n",
      "location/contact/phone_ext 非NA数量: 1\n",
      "patient_data/ipd_info_type 非NA数量: 2625\n",
      "patient_data/ipd_time_frame 非NA数量: 2497\n",
      "patient_data/ipd_access_criteria 非NA数量: 2611\n",
      "clinical_results/outcome_list/outcome/measure/units_analyzed 非NA数量: 1450\n",
      "clinical_results/baseline/measure_list/measure/population 非NA数量: 6162\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/units 非NA数量: 5583\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/scope 非NA数量: 5583\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 5583\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/count_list/count{value} 非NA数量: 5583\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/title 非NA数量: 3204\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/assessment 非NA数量: 641\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement 非NA数量: 3479\n",
      "disposition_first_submitted 非NA数量: 3045\n",
      "disposition_first_submitted_qc 非NA数量: 3045\n",
      "disposition_first_posted 非NA数量: 3045\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/estimate_desc 非NA数量: 2629\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/sub_title{vocab} 非NA数量: 282\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/non_inferiority_desc 非NA数量: 3353\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/other_analysis_desc 非NA数量: 569\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/units 非NA数量: 7140\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/scope 非NA数量: 7140\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 7140\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/count_list/count{value} 非NA数量: 7140\n",
      "clinical_results/baseline/measure_list/measure/units_analyzed 非NA数量: 260\n",
      "patient_data/ipd_url 非NA数量: 2584\n",
      "location/contact_backup/last_name 非NA数量: 0\n",
      "location/contact_backup/phone 非NA数量: 0\n",
      "location/contact_backup/email 非NA数量: 0\n",
      "eligibility/gender_description 非NA数量: 436\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/dispersion_type 非NA数量: 1690\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/dispersion_value 非NA数量: 1690\n",
      "location/investigator/last_name 非NA数量: 1\n",
      "location/investigator/role 非NA数量: 1\n",
      "target_duration 非NA数量: 96\n",
      "pending_results/submission_canceled 非NA数量: 0\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement 非NA数量: 317\n",
      "clinical_results/outcome_list/outcome/posting_date 非NA数量: 436\n",
      "overall_contact/phone_ext 非NA数量: 0\n",
      "location/contact_backup/phone_ext 非NA数量: 0\n",
      "overall_contact_backup/phone_ext 非NA数量: 0\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_upper_limit_na_comment 非NA数量: 24\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/units 非NA数量: 84\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/scope 非NA数量: 84\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 84\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/count_list/count{value} 非NA数量: 84\n"
     ]
    }
   ],
   "source": [
    "task_name = 'adverse-event-rate-prediction'\n",
    "target = pd.read_csv(f'{task_name}/output.csv', index_col=0)\n",
    "All = All_Ori.loc[target.index]\n",
    "target = target['mortality_rate']\n",
    "MI_VAR = []\n",
    "for feature_name in features_category:\n",
    "    print(feature_name, end='  ')\n",
    "    notna = All[feature_name].notna().sum()\n",
    "    print('非NA数量:', notna, end='  ')\n",
    "    var = All[All[feature_name].notna()][feature_name].astype('category').cat.codes.var()\n",
    "    print('Var: ', var, end='  ')\n",
    "    if notna < 5:\n",
    "        mi = 'N/A'\n",
    "        print('MI: ', mi)\n",
    "    else:\n",
    "        mi = NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes.to_numpy().reshape(-1, 1), target.loc[All[All[feature_name].notna()].index])[0]\n",
    "        print('MI: ', mi)\n",
    "\n",
    "    MI_VAR.append((feature_name, notna, var, mi, 'Category'))\n",
    "\n",
    "for feature_name in features_continuous:\n",
    "    print(feature_name, end='  ')\n",
    "    notna = All[feature_name].notna().sum()\n",
    "    print('非NA数量: ', notna, end='  ')\n",
    "    var = All[All[feature_name].notna()][feature_name].var()\n",
    "    print('Var: ', var, end='  ')\n",
    "    if notna < 5:\n",
    "        mi = 'N/A'\n",
    "        print('MI: ', mi)\n",
    "    else:\n",
    "        mi = NMI(All[All[feature_name].notna()][feature_name].to_numpy().reshape(-1, 1), target.loc[All[All[feature_name].notna()].index])[0]\n",
    "        print('MI: ', mi)\n",
    "\n",
    "    MI_VAR.append((feature_name, notna, var, mi, 'Continues'))\n",
    "\n",
    "df = pd.DataFrame(MI_VAR, columns=['feature_name', 'Not NA', 'var', 'MI', 'Category/Continues']).set_index('feature_name')\n",
    "for feature_name in feature_info.index:\n",
    "    if feature_name not in df.index:\n",
    "        df.loc[feature_name] = (All[feature_name].notna().sum(), None, None, None)\n",
    "        print(feature_name, '非NA数量:', All[feature_name].notna().sum())\n",
    "        \n",
    "df.to_csv(f'{task_name}/mortality_rate_MI.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## participant-dropout-prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import KBinsDiscretizer\n",
    "# target = pd.read_csv('trial-duration-prediction/output.csv')['time_day']\n",
    "target = pd.read_csv('participant-flow-prediction/output.csv', index_col=0)['droupout_rate']\n",
    "# 将连续变量离散化\n",
    "# est = KBinsDiscretizer(n_bins=10, encode='ordinal', strategy='uniform')\n",
    "# continuous_target_discrete = est.fit_transform(target.values.reshape(-1, 1)).flatten()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "required_header/download_date  非NA数量: 62058  Var:  0.0  MI:  0.00024888797725131084\n",
      "required_header/link_text  非NA数量: 62058  Var:  0.0  MI:  0.0011895647223001404\n",
      "sponsors/lead_sponsor/agency_class  非NA数量: 62058  Var:  1.0303651724591845  MI:  0.05607231097560561\n",
      "overall_status  非NA数量: 62058  Var:  1.1652013161966617  MI:  0.06296578200516612\n",
      "phase  非NA数量: 42786  Var:  2.5043924114216587  MI:  0.06881710433308452\n",
      "study_type  非NA数量: 62058  Var:  0.05945918507760948  MI:  0.004158037126480885\n",
      "has_expanded_access  非NA数量: 61574  Var:  0.006213644518638787  MI:  0.003160143348626576\n",
      "study_design_info/intervention_model  非NA数量: 58441  Var:  1.3431203429385206  MI:  0.07196649034151381\n",
      "study_design_info/primary_purpose  非NA数量: 57781  Var:  4.2723416587077585  MI:  0.026069729074326364\n",
      "eligibility/gender  非NA数量: 62058  Var:  0.2137329888626816  MI:  0.002332337215137592\n",
      "eligibility/healthy_volunteers  非NA数量: 61964  Var:  0.15884854238204024  MI:  0.02830828895978721\n",
      "study_first_posted{type}  非NA数量: 62058  Var:  0.16172875554051255  MI:  0.004569981715079052\n",
      "last_update_posted{type}  非NA数量: 62058  Var:  0.1981750296233449  MI:  0.0021196254907507317\n",
      "study_design_info/observational_model  非NA数量: 3274  Var:  1.6558995770918497  MI:  0.024305941087312988\n",
      "study_design_info/time_perspective  非NA数量: 3402  Var:  0.5383422865045917  MI:  0.06180613215589714\n",
      "oversight_info/has_dmc  非NA数量: 54777  Var:  0.24380630592348276  MI:  0.00620540158542271\n",
      "completion_date{type}  非NA数量: 61343  Var:  0.017091725155194613  MI:  0.0022156138849793727\n",
      "primary_completion_date{type}  非NA数量: 61983  Var:  0.0004676588648889689  MI:  0\n",
      "enrollment{type}  非NA数量: 62055  Var:  4.834265264717274e-05  MI:  0.0026794621571677446\n",
      "biospec_retention  非NA数量: 507  Var:  0.4068651526845506  MI:  0\n",
      "eligibility/sampling_method  非NA数量: 3438  Var:  0.18921633193713894  MI:  0.012148916782269481\n",
      "responsible_party/responsible_party_type  非NA数量: 60035  Var:  0.2330254437118763  MI:  0.02912471151133955\n",
      "last_known_status  非NA数量: 102  Var:  0.019413706076489998  MI:  0.004693915379941016\n",
      "study_design_info/allocation  非NA数量: 43659  Var:  0.11957255350600594  MI:  0.023983633499975454\n",
      "start_date{type}  非NA数量: 28234  Var:  0.0  MI:  0.0012557920866820282\n",
      "results_first_posted{type}  非NA数量: 62058  Var:  0.23781995278456877  MI:  0.001717576213435379\n",
      "clinical_results/baseline/analyzed_list/analyzed/scope  非NA数量: 62058  Var:  0.00500235916788075  MI:  0.002082408974014349\n",
      "clinical_results/reported_events/serious_events/default_assessment  非NA数量: 32136  Var:  0.19805065189143434  MI:  0.00681091707551662\n",
      "clinical_results/reported_events/other_events/default_assessment  非NA数量: 41017  Var:  0.1989987154136792  MI:  0.0018355795312161582\n",
      "clinical_results/certain_agreements/pi_employee  非NA数量: 62058  Var:  0.19255707504195133  MI:  0.029209009076027304\n",
      "patient_data/sharing_ipd  非NA数量: 23253  Var:  0.8137114490216814  MI:  0.03755735458113607\n",
      "oversight_info/is_fda_regulated_drug  非NA数量: 19441  Var:  0.24293525646765368  MI:  0.023802727820454006\n",
      "oversight_info/is_fda_regulated_device  非NA数量: 19417  Var:  0.1422636680412432  MI:  0.004944801062175319\n",
      "oversight_info/is_us_export  非NA数量: 5706  Var:  0.23353863101497166  MI:  0.009326569730020662\n",
      "disposition_first_posted{type}  非NA数量: 5507  Var:  0.24456889428644493  MI:  0.0024641361300439257\n",
      "eligibility/gender_based  非NA数量: 758  Var:  0.0  MI:  0\n",
      "expanded_access_info/expanded_access_type_intermediate  非NA数量: 0  Var:  nan  MI:  N/A\n",
      "expanded_access_info/expanded_access_type_treatment  非NA数量: 0  Var:  nan  MI:  N/A\n",
      "oversight_info/is_unapproved_device  非NA数量: 702  Var:  0.0  MI:  0.02718406322167799\n",
      "expanded_access_info/expanded_access_type_individual  非NA数量: 0  Var:  nan  MI:  N/A\n",
      "oversight_info/is_ppsd  非NA数量: 2  Var:  0.0  MI:  N/A\n",
      "eligibility/minimum_age  非NA数量: 59387  Var:  106.39352463222828  MI:  0.0180120772995771\n",
      "eligibility/maximum_age  非NA数量: 29515  Var:  746.1156671009272  MI:  0.03517785441611743\n",
      "number_of_arms  非NA数量: 58230  Var:  2.447574707897293  MI:  0.0397545830689201\n",
      "required_header/url 非NA数量: 62058\n",
      "id_info/org_study_id 非NA数量: 62058\n",
      "id_info/secondary_id 非NA数量: 23468\n",
      "id_info/nct_id 非NA数量: 62058\n",
      "brief_title 非NA数量: 62058\n",
      "sponsors/lead_sponsor/agency 非NA数量: 62058\n",
      "source 非NA数量: 62058\n",
      "brief_summary/textblock 非NA数量: 62058\n",
      "detailed_description/textblock 非NA数量: 40396\n",
      "study_design_info/masking 非NA数量: 58509\n",
      "condition 非NA数量: 62058\n",
      "intervention/intervention_type 非NA数量: 60736\n",
      "intervention/intervention_name 非NA数量: 60736\n",
      "eligibility/criteria/textblock 非NA数量: 62055\n",
      "location/facility/name 非NA数量: 54530\n",
      "location/facility/address/city 非NA数量: 57849\n",
      "location/facility/address/state 非NA数量: 50080\n",
      "location/facility/address/country 非NA数量: 57849\n",
      "location_countries/country 非NA数量: 57849\n",
      "verification_date 非NA数量: 62058\n",
      "study_first_submitted 非NA数量: 62058\n",
      "study_first_submitted_qc 非NA数量: 62058\n",
      "study_first_posted 非NA数量: 62058\n",
      "last_update_submitted 非NA数量: 62058\n",
      "last_update_submitted_qc 非NA数量: 62058\n",
      "last_update_posted 非NA数量: 62058\n",
      "condition_browse/mesh_term 非NA数量: 54135\n",
      "intervention_browse/mesh_term 非NA数量: 35656\n",
      "icdcode 非NA数量: 62058\n",
      "sponsors/collaborator/agency 非NA数量: 22132\n",
      "sponsors/collaborator/agency_class 非NA数量: 22132\n",
      "location/facility/address/zip 非NA数量: 51890\n",
      "keyword 非NA数量: 40347\n",
      "official_title 非NA数量: 61283\n",
      "why_stopped 非NA数量: 8278\n",
      "start_date 非NA数量: 62041\n",
      "completion_date 非NA数量: 61364\n",
      "primary_completion_date 非NA数量: 61983\n",
      "primary_outcome/measure 非NA数量: 62058\n",
      "primary_outcome/time_frame 非NA数量: 62051\n",
      "secondary_outcome/measure 非NA数量: 49169\n",
      "secondary_outcome/time_frame 非NA数量: 49163\n",
      "number_of_groups 非NA数量: 2856\n",
      "enrollment 非NA数量: 62055\n",
      "arm_group/arm_group_label 非NA数量: 61086\n",
      "arm_group/description 非NA数量: 55177\n",
      "intervention/description 非NA数量: 56330\n",
      "intervention/arm_group_label 非NA数量: 60130\n",
      "intervention/other_name 非NA数量: 32366\n",
      "biospec_descr/textblock 非NA数量: 507\n",
      "eligibility/study_pop/textblock 非NA数量: 3438\n",
      "overall_official/last_name 非NA数量: 55368\n",
      "overall_official/role 非NA数量: 55363\n",
      "overall_official/affiliation 非NA数量: 55235\n",
      "overall_contact/last_name 非NA数量: 3\n",
      "overall_contact/phone 非NA数量: 3\n",
      "overall_contact/email 非NA数量: 3\n",
      "location/status 非NA数量: 3\n",
      "arm_group/arm_group_type 非NA数量: 58230\n",
      "link/url 非NA数量: 11482\n",
      "link/description 非NA数量: 11482\n",
      "results_reference/citation 非NA数量: 7442\n",
      "results_reference/PMID 非NA数量: 7152\n",
      "reference/citation 非NA数量: 8775\n",
      "reference/PMID 非NA数量: 8555\n",
      "study_design_info/masking_description 非NA数量: 2631\n",
      "primary_outcome/description 非NA数量: 58801\n",
      "secondary_outcome/description 非NA数量: 46874\n",
      "responsible_party/investigator_affiliation 非NA数量: 16200\n",
      "responsible_party/investigator_full_name 非NA数量: 16199\n",
      "responsible_party/investigator_title 非NA数量: 16200\n",
      "acronym 非NA数量: 14066\n",
      "results_first_submitted 非NA数量: 62058\n",
      "results_first_submitted_qc 非NA数量: 62058\n",
      "results_first_posted 非NA数量: 62058\n",
      "provided_document_section/provided_document/document_type 非NA数量: 23911\n",
      "provided_document_section/provided_document/document_has_protocol 非NA数量: 23911\n",
      "provided_document_section/provided_document/document_has_icf 非NA数量: 23911\n",
      "provided_document_section/provided_document/document_has_sap 非NA数量: 23911\n",
      "provided_document_section/provided_document/document_date 非NA数量: 23911\n",
      "provided_document_section/provided_document/document_url 非NA数量: 23911\n",
      "clinical_results/participant_flow/group_list/group{group_id} 非NA数量: 62058\n",
      "clinical_results/participant_flow/group_list/group/title 非NA数量: 62058\n",
      "clinical_results/participant_flow/group_list/group/description 非NA数量: 61578\n",
      "clinical_results/participant_flow/period_list/period/title 非NA数量: 62058\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/title 非NA数量: 62058\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/participants_list/participants{group_id} 非NA数量: 62058\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/participants_list/participants{count} 非NA数量: 62058\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/participants_list/participants 非NA数量: 8166\n",
      "clinical_results/baseline/population 非NA数量: 19717\n",
      "clinical_results/baseline/group_list/group{group_id} 非NA数量: 62058\n",
      "clinical_results/baseline/group_list/group/title 非NA数量: 62058\n",
      "clinical_results/baseline/group_list/group/description 非NA数量: 61624\n",
      "clinical_results/baseline/analyzed_list/analyzed/units 非NA数量: 62058\n",
      "clinical_results/baseline/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 62058\n",
      "clinical_results/baseline/analyzed_list/analyzed/count_list/count{value} 非NA数量: 62058\n",
      "clinical_results/baseline/measure_list/measure/title 非NA数量: 62058\n",
      "clinical_results/baseline/measure_list/measure/units 非NA数量: 62017\n",
      "clinical_results/baseline/measure_list/measure/param 非NA数量: 61934\n",
      "clinical_results/baseline/measure_list/measure/dispersion 非NA数量: 52841\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{group_id} 非NA数量: 61913\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{value} 非NA数量: 61913\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{spread} 非NA数量: 43155\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/title 非NA数量: 61126\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/title 非NA数量: 43904\n",
      "clinical_results/outcome_list/outcome/type 非NA数量: 62058\n",
      "clinical_results/outcome_list/outcome/title 非NA数量: 62058\n",
      "clinical_results/outcome_list/outcome/description 非NA数量: 60186\n",
      "clinical_results/outcome_list/outcome/time_frame 非NA数量: 62053\n",
      "clinical_results/outcome_list/outcome/population 非NA数量: 47622\n",
      "clinical_results/outcome_list/outcome/group_list/group{group_id} 非NA数量: 62058\n",
      "clinical_results/outcome_list/outcome/group_list/group/title 非NA数量: 62058\n",
      "clinical_results/outcome_list/outcome/group_list/group/description 非NA数量: 61392\n",
      "clinical_results/outcome_list/outcome/measure/title 非NA数量: 62058\n",
      "clinical_results/outcome_list/outcome/measure/description 非NA数量: 60151\n",
      "clinical_results/outcome_list/outcome/measure/population 非NA数量: 47622\n",
      "clinical_results/outcome_list/outcome/measure/units 非NA数量: 60221\n",
      "clinical_results/outcome_list/outcome/measure/param 非NA数量: 60222\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/units 非NA数量: 62058\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/scope 非NA数量: 62058\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 62058\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/count_list/count{value} 非NA数量: 62058\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/title 非NA数量: 31309\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{group_id} 非NA数量: 60186\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{value} 非NA数量: 60171\n",
      "clinical_results/reported_events/time_frame 非NA数量: 46601\n",
      "clinical_results/reported_events/desc 非NA数量: 25900\n",
      "clinical_results/reported_events/group_list/group{group_id} 非NA数量: 61855\n",
      "clinical_results/reported_events/group_list/group/title 非NA数量: 61844\n",
      "clinical_results/reported_events/group_list/group/description 非NA数量: 60795\n",
      "clinical_results/reported_events/serious_events/category_list/category/title 非NA数量: 61849\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/sub_title 非NA数量: 61849\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{group_id} 非NA数量: 61849\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{subjects_affected} 非NA数量: 61849\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{subjects_at_risk} 非NA数量: 61841\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/description 非NA数量: 5542\n",
      "clinical_results/reported_events/other_events/frequency_threshold 非NA数量: 61849\n",
      "clinical_results/reported_events/other_events/category_list/category/title 非NA数量: 61849\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/sub_title 非NA数量: 61849\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{group_id} 非NA数量: 61849\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{subjects_affected} 非NA数量: 61849\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{subjects_at_risk} 非NA数量: 61845\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/description 非NA数量: 6226\n",
      "clinical_results/certain_agreements/restrictive_agreement 非NA数量: 60844\n",
      "clinical_results/point_of_contact/name_or_title 非NA数量: 62058\n",
      "clinical_results/point_of_contact/organization 非NA数量: 62058\n",
      "clinical_results/point_of_contact/phone 非NA数量: 58469\n",
      "clinical_results/point_of_contact/email 非NA数量: 57227\n",
      "clinical_results/participant_flow/recruitment_details 非NA数量: 25513\n",
      "clinical_results/reported_events/serious_events/default_vocab 非NA数量: 24361\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{events} 非NA数量: 18844\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/sub_title{vocab} 非NA数量: 692\n",
      "clinical_results/reported_events/other_events/default_vocab 非NA数量: 28884\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{events} 非NA数量: 23989\n",
      "clinical_results/participant_flow/pre_assignment_details 非NA数量: 23837\n",
      "clinical_results/participant_flow/period_list/period/drop_withdraw_reason_list/drop_withdraw_reason/title 非NA数量: 38083\n",
      "clinical_results/participant_flow/period_list/period/drop_withdraw_reason_list/drop_withdraw_reason/participants_list/participants{group_id} 非NA数量: 38083\n",
      "clinical_results/participant_flow/period_list/period/drop_withdraw_reason_list/drop_withdraw_reason/participants_list/participants{count} 非NA数量: 38083\n",
      "id_info/nct_alias 非NA数量: 1108\n",
      "responsible_party/name_title 非NA数量: 1792\n",
      "responsible_party/organization 非NA数量: 1798\n",
      "location/contact/last_name 非NA数量: 1\n",
      "location/contact/phone 非NA数量: 1\n",
      "location/contact/email 非NA数量: 1\n",
      "removed_countries/country 非NA数量: 6568\n",
      "clinical_results/outcome_list/outcome/measure/dispersion 非NA数量: 48995\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{spread} 非NA数量: 34975\n",
      "clinical_results/limitations_and_caveats 非NA数量: 12994\n",
      "other_outcome/measure 非NA数量: 4837\n",
      "other_outcome/time_frame 非NA数量: 4836\n",
      "pending_results/submitted 非NA数量: 0\n",
      "pending_results/returned 非NA数量: 0\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/group_id_list/group_id 非NA数量: 21424\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/groups_desc 非NA数量: 11407\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/non_inferiority_type 非NA数量: 21424\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/p_value 非NA数量: 18286\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/p_value_desc 非NA数量: 6767\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/method 非NA数量: 18286\n",
      "clinical_results/baseline/measure_list/measure/description 非NA数量: 12821\n",
      "study_design_info/intervention_model_description 非NA数量: 4533\n",
      "study_docs/study_doc/doc_id 非NA数量: 730\n",
      "study_docs/study_doc/doc_type 非NA数量: 1007\n",
      "study_docs/study_doc/doc_url 非NA数量: 1007\n",
      "study_docs/study_doc/doc_comment 非NA数量: 754\n",
      "patient_data/ipd_description 非NA数量: 8725\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{lower_limit} 非NA数量: 24642\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{upper_limit} 非NA数量: 24548\n",
      "other_outcome/description 非NA数量: 4524\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{lower_limit} 非NA数量: 11579\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{upper_limit} 非NA数量: 11559\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/param_type 非NA数量: 14195\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/param_value 非NA数量: 14195\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_percent 非NA数量: 13091\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_n_sides 非NA数量: 13034\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_lower_limit 非NA数量: 12625\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_upper_limit 非NA数量: 12641\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/method_desc 非NA数量: 5379\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/assessment 非NA数量: 729\n",
      "overall_contact_backup/last_name 非NA数量: 2\n",
      "overall_contact_backup/phone 非NA数量: 2\n",
      "overall_contact_backup/email 非NA数量: 1\n",
      "location/contact/phone_ext 非NA数量: 1\n",
      "patient_data/ipd_info_type 非NA数量: 2985\n",
      "patient_data/ipd_time_frame 非NA数量: 2876\n",
      "patient_data/ipd_access_criteria 非NA数量: 3001\n",
      "clinical_results/outcome_list/outcome/measure/units_analyzed 非NA数量: 2614\n",
      "clinical_results/baseline/measure_list/measure/population 非NA数量: 7047\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/units 非NA数量: 6385\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/scope 非NA数量: 6385\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 6385\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/count_list/count{value} 非NA数量: 6385\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/title 非NA数量: 3657\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/assessment 非NA数量: 1604\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement 非NA数量: 5931\n",
      "disposition_first_submitted 非NA数量: 5507\n",
      "disposition_first_submitted_qc 非NA数量: 5507\n",
      "disposition_first_posted 非NA数量: 5507\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/estimate_desc 非NA数量: 5406\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/sub_title{vocab} 非NA数量: 846\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/non_inferiority_desc 非NA数量: 5263\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/other_analysis_desc 非NA数量: 634\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/units 非NA数量: 8513\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/scope 非NA数量: 8513\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 8513\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/count_list/count{value} 非NA数量: 8513\n",
      "clinical_results/baseline/measure_list/measure/units_analyzed 非NA数量: 313\n",
      "patient_data/ipd_url 非NA数量: 3395\n",
      "location/contact_backup/last_name 非NA数量: 0\n",
      "location/contact_backup/phone 非NA数量: 0\n",
      "location/contact_backup/email 非NA数量: 0\n",
      "eligibility/gender_description 非NA数量: 464\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/dispersion_type 非NA数量: 3532\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/dispersion_value 非NA数量: 3532\n",
      "location/investigator/last_name 非NA数量: 1\n",
      "location/investigator/role 非NA数量: 1\n",
      "target_duration 非NA数量: 152\n",
      "pending_results/submission_canceled 非NA数量: 0\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement 非NA数量: 604\n",
      "clinical_results/outcome_list/outcome/posting_date 非NA数量: 530\n",
      "overall_contact/phone_ext 非NA数量: 0\n",
      "location/contact_backup/phone_ext 非NA数量: 0\n",
      "overall_contact_backup/phone_ext 非NA数量: 0\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_upper_limit_na_comment 非NA数量: 29\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/units 非NA数量: 102\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/scope 非NA数量: 102\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 102\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/count_list/count{value} 非NA数量: 102\n"
     ]
    }
   ],
   "source": [
    "task_name = 'participant-flow-prediction'\n",
    "target = pd.read_csv(f'{task_name}/output.csv', index_col=0)\n",
    "All = All_Ori.loc[target.index]\n",
    "target = target['droupout_rate']\n",
    "MI_VAR = []\n",
    "for feature_name in features_category:\n",
    "    print(feature_name, end='  ')\n",
    "    notna = All[feature_name].notna().sum()\n",
    "    print('非NA数量:', notna, end='  ')\n",
    "    var = All[All[feature_name].notna()][feature_name].astype('category').cat.codes.var()\n",
    "    print('Var: ', var, end='  ')\n",
    "    if notna < 5:\n",
    "        mi = 'N/A'\n",
    "        print('MI: ', mi)\n",
    "    else:\n",
    "        mi = NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes.to_numpy().reshape(-1, 1), target.loc[All[All[feature_name].notna()].index])[0]\n",
    "        print('MI: ', mi)\n",
    "\n",
    "    MI_VAR.append((feature_name, notna, var, mi, 'Category'))\n",
    "\n",
    "for feature_name in features_continuous:\n",
    "    print(feature_name, end='  ')\n",
    "    notna = All[feature_name].notna().sum()\n",
    "    print('非NA数量:', notna, end='  ')\n",
    "    var = All[All[feature_name].notna()][feature_name].var()\n",
    "    print('Var: ', var, end='  ')\n",
    "    if notna < 5:\n",
    "        mi = 'N/A'\n",
    "        print('MI: ', mi)\n",
    "    else:\n",
    "        mi = NMI(All[All[feature_name].notna()][feature_name].to_numpy().reshape(-1, 1), target.loc[All[All[feature_name].notna()].index])[0]\n",
    "        print('MI: ', mi)\n",
    "\n",
    "    MI_VAR.append((feature_name, notna, var, mi, 'Continues'))\n",
    "\n",
    "df = pd.DataFrame(MI_VAR, columns=['feature_name', 'Not NA', 'var', 'MI', 'Category/Continues']).set_index('feature_name')\n",
    "for feature_name in feature_info.index:\n",
    "    if feature_name not in df.index:\n",
    "        df.loc[feature_name] = (All[feature_name].notna().sum(), None, None, None)\n",
    "        print(feature_name, '非NA数量:', All[feature_name].notna().sum())\n",
    "        \n",
    "df.to_csv(f'{task_name}/participants_dropout_rate_MI.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "required_header/download_date  非NA数量: 62058  NMI:  0.0\n",
      "required_header/link_text  非NA数量: 62058  NMI:  0.0\n",
      "sponsors/lead_sponsor/agency_class  非NA数量: 62058  NMI:  0.009851515106071315\n",
      "overall_status  非NA数量: 62058  NMI:  0.04130900160348425\n",
      "phase  非NA数量: 42786  NMI:  0.013705405790074465\n",
      "study_type  非NA数量: 62058  NMI:  0.002811026538073855\n",
      "has_expanded_access  非NA数量: 61574  NMI:  0.0021548332918209334\n",
      "study_design_info/intervention_model  非NA数量: 58441  NMI:  0.01705521572004618\n",
      "study_design_info/primary_purpose  非NA数量: 57781  NMI:  0.014057163916747478\n",
      "eligibility/gender  非NA数量: 62058  NMI:  0.0006461978030121632\n",
      "eligibility/healthy_volunteers  非NA数量: 61964  NMI:  0.01946112079829883\n",
      "study_first_posted{type}  非NA数量: 62058  NMI:  0.0019778382727979387\n",
      "last_update_posted{type}  非NA数量: 62058  NMI:  0.0006248423236193851\n",
      "study_design_info/observational_model  非NA数量: 3274  NMI:  0.013951129236524073\n",
      "study_design_info/time_perspective  非NA数量: 3402  NMI:  0.035008285293421045\n",
      "oversight_info/has_dmc  非NA数量: 54777  NMI:  0.0026859269807584765\n",
      "completion_date{type}  非NA数量: 61343  NMI:  0.004393009912790445\n",
      "primary_completion_date{type}  非NA数量: 61983  NMI:  0.00015042999941397946\n",
      "enrollment{type}  非NA数量: 62055  NMI:  9.039659364851382e-05\n",
      "biospec_retention  非NA数量: 507  NMI:  0.01600584368931936\n",
      "eligibility/sampling_method  非NA数量: 3438  NMI:  0.0008354964214184529\n",
      "responsible_party/responsible_party_type  非NA数量: 60035  NMI:  0.006365318184808411\n",
      "last_known_status  非NA数量: 102  NMI:  0.025069007539954297\n",
      "study_design_info/allocation  非NA数量: 43659  NMI:  0.010456949486698994\n",
      "start_date{type}  非NA数量: 28234  NMI:  0.0\n",
      "results_first_posted{type}  非NA数量: 62058  NMI:  0.0008256070070487956\n",
      "clinical_results/baseline/analyzed_list/analyzed/scope  非NA数量: 62058  NMI:  0.0002814640973977265\n",
      "clinical_results/reported_events/serious_events/default_assessment  非NA数量: 32136  NMI:  0.001065305493684422\n",
      "clinical_results/reported_events/other_events/default_assessment  非NA数量: 41017  NMI:  0.0009761563203467776\n",
      "clinical_results/certain_agreements/pi_employee  非NA数量: 62058  NMI:  0.0037985671931423342\n",
      "patient_data/sharing_ipd  非NA数量: 23253  NMI:  0.005962271592283877\n",
      "oversight_info/is_fda_regulated_drug  非NA数量: 19441  NMI:  0.01839424392833271\n",
      "oversight_info/is_fda_regulated_device  非NA数量: 19417  NMI:  0.006822680675844718\n",
      "oversight_info/is_us_export  非NA数量: 5706  NMI:  0.002640244609352338\n",
      "disposition_first_posted{type}  非NA数量: 5507  NMI:  0.0008318265734443812\n",
      "eligibility/gender_based  非NA数量: 758  NMI:  0.0\n",
      "expanded_access_info/expanded_access_type_intermediate  非NA数量: 0  NMI:  1.0\n",
      "expanded_access_info/expanded_access_type_treatment  非NA数量: 0  NMI:  1.0\n",
      "oversight_info/is_unapproved_device  非NA数量: 702  NMI:  0.0\n",
      "expanded_access_info/expanded_access_type_individual  非NA数量: 0  NMI:  1.0\n",
      "oversight_info/is_ppsd  非NA数量: 2  NMI:  1.0\n"
     ]
    }
   ],
   "source": [
    "for feature_name in features_category:\n",
    "    print(feature_name, end='  ')\n",
    "    print('非NA数量:', All[feature_name].notna().sum(), end='  ')\n",
    "    print('NMI: ', NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes, continuous_target_discrete[All[feature_name].notna()]))\n",
    "    # sns.countplot(data=All, y=feature_name, hue=feature_name)\n",
    "    # plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## trial-approval-prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "required_header/download_date  非NA数量: 291395  Var:  0.0\n",
      "MI:  0required_header/link_text  非NA数量: 291395  Var:  0.0\n",
      "MI:  0.0006303933207503221sponsors/lead_sponsor/agency_class  非NA数量: 291395  Var:  0.842102800235118\n",
      "MI:  0.0008946939130680676overall_status  非NA数量: 291395  Var:  0.08689541585340167\n",
      "MI:  0.3148296521184868phase  非NA数量: 129333  Var:  3.4112489941107147\n",
      "MI:  0.00194721257915198study_type  非NA数量: 291395  Var:  0.19973192086091807\n",
      "MI:  0.002471510950424971has_expanded_access  非NA数量: 287853  Var:  0.0018724448082320375\n",
      "MI:  0.003378463231458184study_design_info/intervention_model  非NA数量: 226252  Var:  1.415968269231446\n",
      "MI:  0.004959752022353214study_design_info/primary_purpose  非NA数量: 225623  Var:  6.812449503285627\n",
      "MI:  0.001522209896172022eligibility/gender  非NA数量: 291382  Var:  0.2585974998467095\n",
      "MI:  0eligibility/healthy_volunteers  非NA数量: 286364  Var:  0.20612840355174458\n",
      "MI:  0.00799415500737144study_first_posted{type}  非NA数量: 291395  Var:  0.2283469281422402\n",
      "MI:  0.00225500751998009last_update_posted{type}  非NA数量: 291395  Var:  0.22922560976890943\n",
      "MI:  0study_design_info/observational_model  非NA数量: 54856  Var:  3.8934582490779683\n",
      "MI:  0.007099054581824582study_design_info/time_perspective  非NA数量: 57208  Var:  0.7370267707947696\n",
      "MI:  0.008321027688771743oversight_info/has_dmc  非NA数量: 238622  Var:  0.2245195115364128\n",
      "MI:  0.001738280133565695completion_date{type}  非NA数量: 277114  Var:  0.0014917434274965603\n",
      "MI:  0.005918649102089724primary_completion_date{type}  非NA数量: 276151  Var:  0.001251372382959577\n",
      "MI:  0.0012557077638932768enrollment{type}  非NA数量: 278493  Var:  0.03548732532016845\n",
      "MI:  8.600113305146095e-05biospec_retention  非NA数量: 10824  Var:  0.3106854301892871\n",
      "MI:  0eligibility/sampling_method  非NA数量: 56306  Var:  0.20010996214050816\n",
      "MI:  0.0009103702210779829responsible_party/responsible_party_type  非NA数量: 257932  Var:  0.30055206637443327\n",
      "MI:  0.0020234657503426234last_known_status  非NA数量: 0  Var:  nan\n",
      "MI:  0\n",
      "study_design_info/allocation  非NA数量: 180048  Var:  0.12119236889141893\n",
      "MI:  0.0026095725489225785start_date{type}  非NA数量: 142935  Var:  0.0\n",
      "MI:  0results_first_posted{type}  非NA数量: 61202  Var:  0.23885230526695084\n",
      "MI:  0.0006613952917797761clinical_results/baseline/analyzed_list/analyzed/scope  非NA数量: 61202  Var:  0.005314502233852594\n",
      "MI:  0clinical_results/reported_events/serious_events/default_assessment  非NA数量: 31184  Var:  0.19882282719332392\n",
      "MI:  5.764857194634487e-05clinical_results/reported_events/other_events/default_assessment  非NA数量: 40056  Var:  0.1996447658336374\n",
      "MI:  0clinical_results/certain_agreements/pi_employee  非NA数量: 61202  Var:  0.19280619430483825\n",
      "MI:  0.005984517989658933patient_data/sharing_ipd  非NA数量: 110741  Var:  0.578887561826218\n",
      "MI:  0oversight_info/is_fda_regulated_drug  非NA数量: 114352  Var:  0.14828095432058874\n",
      "MI:  0.020727193371564923oversight_info/is_fda_regulated_device  非NA数量: 114358  Var:  0.05527165958582904\n",
      "MI:  0.002792022716996634oversight_info/is_us_export  非NA数量: 16330  Var:  0.21204408793059817\n",
      "MI:  0disposition_first_posted{type}  非NA数量: 8613  Var:  0.2397189700551489\n",
      "MI:  0.008245970200905894eligibility/gender_based  非NA数量: 6062  Var:  0.0\n",
      "MI:  0.008094964751713896expanded_access_info/expanded_access_type_intermediate  非NA数量: 0  Var:  nan\n",
      "MI:  0\n",
      "expanded_access_info/expanded_access_type_treatment  非NA数量: 0  Var:  nan\n",
      "MI:  0\n",
      "oversight_info/is_unapproved_device  非NA数量: 1621  Var:  0.0\n",
      "MI:  0expanded_access_info/expanded_access_type_individual  非NA数量: 0  Var:  nan\n",
      "MI:  0\n",
      "oversight_info/is_ppsd  非NA数量: 4  Var:  0.0\n",
      "MI:  0"
     ]
    }
   ],
   "source": [
    "task_name = 'trial-approval-prediction'\n",
    "target = pd.read_csv(f'{task_name}/output.csv', index_col=0)\n",
    "All = All_Ori.loc[target.index]\n",
    "target = target['label']\n",
    "MI_VAR = []\n",
    "for feature_name in features_category:\n",
    "    print(feature_name, end='  ')\n",
    "    notna = All[feature_name].notna().sum()\n",
    "    print('非NA数量:', notna, end='  ')\n",
    "    var = All[All[feature_name].notna()][feature_name].astype('category').cat.codes.var()\n",
    "    print('Var: ', var, end='  ')\n",
    "    if notna < 5:\n",
    "        mi = 'N/A'\n",
    "        print('MI: ', mi)\n",
    "    else:\n",
    "        mi = NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes.to_numpy().reshape(-1, 1), target.loc[All[All[feature_name].notna()].index])[0]\n",
    "        print('MI: ', mi)\n",
    "\n",
    "    MI_VAR.append((feature_name, notna, var, mi, 'Category'))\n",
    "\n",
    "for feature_name in features_continuous:\n",
    "    print(feature_name, end='  ')\n",
    "    notna = All[feature_name].notna().sum()\n",
    "    print('非NA数量:', notna, end='  ')\n",
    "    var = All[All[feature_name].notna()][feature_name].var()\n",
    "    print('Var: ', var, end='  ')\n",
    "    if notna < 5:\n",
    "        mi = 'N/A'\n",
    "        print('MI: ', mi)\n",
    "    else:\n",
    "        mi = NMI(All[All[feature_name].notna()][feature_name].to_numpy().reshape(-1, 1), target.loc[All[All[feature_name].notna()].index])[0]\n",
    "        print('MI: ', mi)\n",
    "\n",
    "    MI_VAR.append((feature_name, notna, var, mi, 'Continues'))\n",
    "\n",
    "df = pd.DataFrame(MI_VAR, columns=['feature_name', 'Not NA', 'var', 'MI', 'Category/Continues']).set_index('feature_name')\n",
    "for feature_name in feature_info.index:\n",
    "    if feature_name not in df.index:\n",
    "        df.loc[feature_name] = (All[feature_name].notna().sum(), None, None, None)\n",
    "        print(feature_name, '非NA数量:', All[feature_name].notna().sum())\n",
    "\n",
    "df.to_csv(f'{task_name}/{task_name}_MI.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## trial-duration-prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "required_header/url 非NA数量: 462610\n",
      "id_info/org_study_id 非NA数量: 462552\n",
      "id_info/secondary_id 非NA数量: 91167\n",
      "id_info/nct_id 非NA数量: 462610\n",
      "brief_title 非NA数量: 462610\n",
      "sponsors/lead_sponsor/agency 非NA数量: 462610\n",
      "source 非NA数量: 462610\n",
      "brief_summary/textblock 非NA数量: 462610\n",
      "detailed_description/textblock 非NA数量: 313245\n",
      "study_design_info/masking 非NA数量: 354886\n",
      "condition 非NA数量: 462607\n",
      "intervention/intervention_type 非NA数量: 419415\n",
      "intervention/intervention_name 非NA数量: 419415\n",
      "eligibility/criteria/textblock 非NA数量: 462575\n",
      "location/facility/name 非NA数量: 407984\n",
      "location/facility/address/city 非NA数量: 415930\n",
      "location/facility/address/state 非NA数量: 272822\n",
      "location/facility/address/country 非NA数量: 415930\n",
      "location_countries/country 非NA数量: 415930\n",
      "verification_date 非NA数量: 462610\n",
      "study_first_submitted 非NA数量: 462610\n",
      "study_first_submitted_qc 非NA数量: 462610\n",
      "study_first_posted 非NA数量: 462610\n",
      "last_update_submitted 非NA数量: 462610\n",
      "last_update_submitted_qc 非NA数量: 462610\n",
      "last_update_posted 非NA数量: 462610\n",
      "condition_browse/mesh_term 非NA数量: 382993\n",
      "intervention_browse/mesh_term 非NA数量: 147531\n",
      "icdcode 非NA数量: 462607\n",
      "sponsors/collaborator/agency 非NA数量: 153144\n",
      "sponsors/collaborator/agency_class 非NA数量: 153144\n",
      "location/facility/address/zip 非NA数量: 355316\n",
      "keyword 非NA数量: 288902\n",
      "official_title 非NA数量: 454710\n",
      "why_stopped 非NA数量: 35883\n",
      "start_date 非NA数量: 462610\n",
      "completion_date 非NA数量: 453734\n",
      "primary_completion_date 非NA数量: 456450\n",
      "primary_outcome/measure 非NA数量: 454215\n",
      "primary_outcome/time_frame 非NA数量: 446267\n",
      "secondary_outcome/measure 非NA数量: 343083\n",
      "secondary_outcome/time_frame 非NA数量: 336632\n",
      "number_of_groups 非NA数量: 69212\n",
      "enrollment 非NA数量: 459958\n",
      "arm_group/arm_group_label 非NA数量: 412059\n",
      "arm_group/description 非NA数量: 364024\n",
      "intervention/description 非NA数量: 383102\n",
      "intervention/arm_group_label 非NA数量: 390025\n",
      "intervention/other_name 非NA数量: 131809\n",
      "biospec_descr/textblock 非NA数量: 22667\n",
      "eligibility/study_pop/textblock 非NA数量: 103386\n",
      "overall_official/last_name 非NA数量: 352301\n",
      "overall_official/role 非NA数量: 351790\n",
      "overall_official/affiliation 非NA数量: 351442\n",
      "overall_contact/last_name 非NA数量: 121666\n",
      "overall_contact/phone 非NA数量: 119389\n",
      "overall_contact/email 非NA数量: 120960\n",
      "location/status 非NA数量: 100498\n",
      "arm_group/arm_group_type 非NA数量: 342847\n",
      "link/url 非NA数量: 46859\n",
      "link/description 非NA数量: 46859\n",
      "results_reference/citation 非NA数量: 31791\n",
      "results_reference/PMID 非NA数量: 30646\n",
      "reference/citation 非NA数量: 73200\n",
      "reference/PMID 非NA数量: 71759\n",
      "study_design_info/masking_description 非NA数量: 32006\n",
      "primary_outcome/description 非NA数量: 356775\n",
      "secondary_outcome/description 非NA数量: 273603\n",
      "responsible_party/investigator_affiliation 非NA数量: 186997\n",
      "responsible_party/investigator_full_name 非NA数量: 186983\n",
      "responsible_party/investigator_title 非NA数量: 186989\n",
      "acronym 非NA数量: 131396\n",
      "results_first_submitted 非NA数量: 62022\n",
      "results_first_submitted_qc 非NA数量: 62022\n",
      "results_first_posted 非NA数量: 62022\n",
      "provided_document_section/provided_document/document_type 非NA数量: 32714\n",
      "provided_document_section/provided_document/document_has_protocol 非NA数量: 32714\n",
      "provided_document_section/provided_document/document_has_icf 非NA数量: 32714\n",
      "provided_document_section/provided_document/document_has_sap 非NA数量: 32714\n",
      "provided_document_section/provided_document/document_date 非NA数量: 32714\n",
      "provided_document_section/provided_document/document_url 非NA数量: 32714\n",
      "clinical_results/participant_flow/group_list/group{group_id} 非NA数量: 62022\n",
      "clinical_results/participant_flow/group_list/group/title 非NA数量: 62022\n",
      "clinical_results/participant_flow/group_list/group/description 非NA数量: 61544\n",
      "clinical_results/participant_flow/period_list/period/title 非NA数量: 62022\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/title 非NA数量: 62022\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/participants_list/participants{group_id} 非NA数量: 62022\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/participants_list/participants{count} 非NA数量: 62022\n",
      "clinical_results/participant_flow/period_list/period/milestone_list/milestone/participants_list/participants 非NA数量: 8209\n",
      "clinical_results/baseline/population 非NA数量: 19897\n",
      "clinical_results/baseline/group_list/group{group_id} 非NA数量: 62022\n",
      "clinical_results/baseline/group_list/group/title 非NA数量: 62022\n",
      "clinical_results/baseline/group_list/group/description 非NA数量: 61592\n",
      "clinical_results/baseline/analyzed_list/analyzed/units 非NA数量: 62022\n",
      "clinical_results/baseline/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 62022\n",
      "clinical_results/baseline/analyzed_list/analyzed/count_list/count{value} 非NA数量: 62022\n",
      "clinical_results/baseline/measure_list/measure/title 非NA数量: 62022\n",
      "clinical_results/baseline/measure_list/measure/units 非NA数量: 61927\n",
      "clinical_results/baseline/measure_list/measure/param 非NA数量: 61731\n",
      "clinical_results/baseline/measure_list/measure/dispersion 非NA数量: 52711\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{group_id} 非NA数量: 61709\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{value} 非NA数量: 61709\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{spread} 非NA数量: 43038\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/title 非NA数量: 61076\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/title 非NA数量: 43854\n",
      "clinical_results/outcome_list/outcome/type 非NA数量: 62022\n",
      "clinical_results/outcome_list/outcome/title 非NA数量: 62022\n",
      "clinical_results/outcome_list/outcome/description 非NA数量: 60100\n",
      "clinical_results/outcome_list/outcome/time_frame 非NA数量: 62015\n",
      "clinical_results/outcome_list/outcome/population 非NA数量: 47651\n",
      "clinical_results/outcome_list/outcome/group_list/group{group_id} 非NA数量: 62022\n",
      "clinical_results/outcome_list/outcome/group_list/group/title 非NA数量: 62022\n",
      "clinical_results/outcome_list/outcome/group_list/group/description 非NA数量: 61363\n",
      "clinical_results/outcome_list/outcome/measure/title 非NA数量: 62022\n",
      "clinical_results/outcome_list/outcome/measure/description 非NA数量: 60065\n",
      "clinical_results/outcome_list/outcome/measure/population 非NA数量: 47651\n",
      "clinical_results/outcome_list/outcome/measure/units 非NA数量: 59994\n",
      "clinical_results/outcome_list/outcome/measure/param 非NA数量: 59996\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/units 非NA数量: 62022\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/scope 非NA数量: 62022\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 62022\n",
      "clinical_results/outcome_list/outcome/measure/analyzed_list/analyzed/count_list/count{value} 非NA数量: 62022\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/title 非NA数量: 31245\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{group_id} 非NA数量: 59959\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{value} 非NA数量: 59944\n",
      "clinical_results/reported_events/time_frame 非NA数量: 46657\n",
      "clinical_results/reported_events/desc 非NA数量: 26054\n",
      "clinical_results/reported_events/group_list/group{group_id} 非NA数量: 61890\n",
      "clinical_results/reported_events/group_list/group/title 非NA数量: 61879\n",
      "clinical_results/reported_events/group_list/group/description 非NA数量: 60838\n",
      "clinical_results/reported_events/serious_events/category_list/category/title 非NA数量: 61884\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/sub_title 非NA数量: 61884\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{group_id} 非NA数量: 61884\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{subjects_affected} 非NA数量: 61884\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{subjects_at_risk} 非NA数量: 61875\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/description 非NA数量: 5545\n",
      "clinical_results/reported_events/other_events/frequency_threshold 非NA数量: 61884\n",
      "clinical_results/reported_events/other_events/category_list/category/title 非NA数量: 61884\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/sub_title 非NA数量: 61884\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{group_id} 非NA数量: 61884\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{subjects_affected} 非NA数量: 61884\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{subjects_at_risk} 非NA数量: 61879\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/description 非NA数量: 6215\n",
      "clinical_results/certain_agreements/restrictive_agreement 非NA数量: 60810\n",
      "clinical_results/point_of_contact/name_or_title 非NA数量: 62022\n",
      "clinical_results/point_of_contact/organization 非NA数量: 62022\n",
      "clinical_results/point_of_contact/phone 非NA数量: 58429\n",
      "clinical_results/point_of_contact/email 非NA数量: 57219\n",
      "clinical_results/participant_flow/recruitment_details 非NA数量: 25543\n",
      "clinical_results/reported_events/serious_events/default_vocab 非NA数量: 24388\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/counts{events} 非NA数量: 18862\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/sub_title{vocab} 非NA数量: 692\n",
      "clinical_results/reported_events/other_events/default_vocab 非NA数量: 28883\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/counts{events} 非NA数量: 23949\n",
      "clinical_results/participant_flow/pre_assignment_details 非NA数量: 23928\n",
      "clinical_results/participant_flow/period_list/period/drop_withdraw_reason_list/drop_withdraw_reason/title 非NA数量: 37996\n",
      "clinical_results/participant_flow/period_list/period/drop_withdraw_reason_list/drop_withdraw_reason/participants_list/participants{group_id} 非NA数量: 37996\n",
      "clinical_results/participant_flow/period_list/period/drop_withdraw_reason_list/drop_withdraw_reason/participants_list/participants{count} 非NA数量: 37996\n",
      "id_info/nct_alias 非NA数量: 2720\n",
      "responsible_party/name_title 非NA数量: 22697\n",
      "responsible_party/organization 非NA数量: 22614\n",
      "location/contact/last_name 非NA数量: 103031\n",
      "location/contact/phone 非NA数量: 83749\n",
      "location/contact/email 非NA数量: 88530\n",
      "removed_countries/country 非NA数量: 14257\n",
      "clinical_results/outcome_list/outcome/measure/dispersion 非NA数量: 48811\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{spread} 非NA数量: 34818\n",
      "clinical_results/limitations_and_caveats 非NA数量: 13098\n",
      "other_outcome/measure 非NA数量: 38131\n",
      "other_outcome/time_frame 非NA数量: 38112\n",
      "pending_results/submitted 非NA数量: 10523\n",
      "pending_results/returned 非NA数量: 8543\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/group_id_list/group_id 非NA数量: 21313\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/groups_desc 非NA数量: 11391\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/non_inferiority_type 非NA数量: 21313\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/p_value 非NA数量: 18262\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/p_value_desc 非NA数量: 6762\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/method 非NA数量: 18262\n",
      "clinical_results/baseline/measure_list/measure/description 非NA数量: 12830\n",
      "study_design_info/intervention_model_description 非NA数量: 62070\n",
      "study_docs/study_doc/doc_id 非NA数量: 1600\n",
      "study_docs/study_doc/doc_type 非NA数量: 2589\n",
      "study_docs/study_doc/doc_url 非NA数量: 2589\n",
      "study_docs/study_doc/doc_comment 非NA数量: 1711\n",
      "patient_data/ipd_description 非NA数量: 55398\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{lower_limit} 非NA数量: 24618\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement{upper_limit} 非NA数量: 24524\n",
      "other_outcome/description 非NA数量: 34192\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{lower_limit} 非NA数量: 11555\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement{upper_limit} 非NA数量: 11535\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/param_type 非NA数量: 14098\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/param_value 非NA数量: 14098\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_percent 非NA数量: 12997\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_n_sides 非NA数量: 13016\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_lower_limit 非NA数量: 12538\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_upper_limit 非NA数量: 12555\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/method_desc 非NA数量: 5370\n",
      "clinical_results/reported_events/serious_events/category_list/category/event_list/event/assessment 非NA数量: 731\n",
      "overall_contact_backup/last_name 非NA数量: 59573\n",
      "overall_contact_backup/phone 非NA数量: 50733\n",
      "overall_contact_backup/email 非NA数量: 56332\n",
      "location/contact/phone_ext 非NA数量: 12322\n",
      "patient_data/ipd_info_type 非NA数量: 20943\n",
      "patient_data/ipd_time_frame 非NA数量: 20276\n",
      "patient_data/ipd_access_criteria 非NA数量: 18519\n",
      "clinical_results/outcome_list/outcome/measure/units_analyzed 非NA数量: 2610\n",
      "clinical_results/baseline/measure_list/measure/population 非NA数量: 7075\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/units 非NA数量: 6378\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/scope 非NA数量: 6378\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 6378\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/analyzed_list/analyzed/count_list/count{value} 非NA数量: 6378\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/title 非NA数量: 3660\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/assessment 非NA数量: 1603\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/category_list/category/measurement_list/measurement 非NA数量: 5961\n",
      "disposition_first_submitted 非NA数量: 8798\n",
      "disposition_first_submitted_qc 非NA数量: 8769\n",
      "disposition_first_posted 非NA数量: 8769\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/estimate_desc 非NA数量: 5331\n",
      "clinical_results/reported_events/other_events/category_list/category/event_list/event/sub_title{vocab} 非NA数量: 844\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/non_inferiority_desc 非NA数量: 5178\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/other_analysis_desc 非NA数量: 634\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/units 非NA数量: 8533\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/scope 非NA数量: 8533\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 8533\n",
      "clinical_results/outcome_list/outcome/measure/class_list/class/analyzed_list/analyzed/count_list/count{value} 非NA数量: 8533\n",
      "clinical_results/baseline/measure_list/measure/units_analyzed 非NA数量: 331\n",
      "patient_data/ipd_url 非NA数量: 9183\n",
      "location/contact_backup/last_name 非NA数量: 27963\n",
      "location/contact_backup/phone 非NA数量: 23113\n",
      "location/contact_backup/email 非NA数量: 26274\n",
      "eligibility/gender_description 非NA数量: 8506\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/dispersion_type 非NA数量: 3527\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/dispersion_value 非NA数量: 3527\n",
      "location/investigator/last_name 非NA数量: 41120\n",
      "location/investigator/role 非NA数量: 41120\n",
      "target_duration 非NA数量: 11374\n",
      "pending_results/submission_canceled 非NA数量: 854\n",
      "clinical_results/baseline/measure_list/measure/class_list/class/category_list/category/measurement_list/measurement 非NA数量: 619\n",
      "clinical_results/outcome_list/outcome/posting_date 非NA数量: 531\n",
      "overall_contact/phone_ext 非NA数量: 15862\n",
      "location/contact_backup/phone_ext 非NA数量: 3692\n",
      "overall_contact_backup/phone_ext 非NA数量: 7326\n",
      "clinical_results/outcome_list/outcome/analysis_list/analysis/ci_upper_limit_na_comment 非NA数量: 29\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/units 非NA数量: 107\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/scope 非NA数量: 107\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/count_list/count{group_id} 非NA数量: 107\n",
      "clinical_results/baseline/measure_list/measure/analyzed_list/analyzed/count_list/count{value} 非NA数量: 107\n"
     ]
    }
   ],
   "source": [
    "# task_name = 'trial-duration-prediction'\n",
    "# target = pd.read_csv(f'{task_name}/output.csv', index_col=0)\n",
    "# All = All_Ori.loc[target.index]\n",
    "# target = target['time_day']\n",
    "# MI_VAR = []\n",
    "# for feature_name in features_category:\n",
    "#     print(feature_name, end='  ')\n",
    "#     notna = All[feature_name].notna().sum()\n",
    "#     print('非NA数量:', notna, end='  ')\n",
    "#     var = All[All[feature_name].notna()][feature_name].astype('category').cat.codes.var()\n",
    "#     print('Var: ', var, end='  ')\n",
    "#     if notna < 5:\n",
    "#         mi = 'N/A'\n",
    "#         print('MI: ', mi)\n",
    "#     else:\n",
    "#         mi = NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes.to_numpy().reshape(-1, 1), target.loc[All[All[feature_name].notna()].index])[0]\n",
    "#         print('MI: ', mi)\n",
    "\n",
    "#     MI_VAR.append((feature_name, notna, var, mi, 'Category'))\n",
    "\n",
    "# for feature_name in features_continuous:\n",
    "#     print(feature_name, end='  ')\n",
    "#     notna = All[feature_name].notna().sum()\n",
    "#     print('非NA数量:', notna, end='  ')\n",
    "#     var = All[All[feature_name].notna()][feature_name].var()\n",
    "#     print('Var: ', var, end='  ')\n",
    "#     if notna < 5:\n",
    "#         mi = 'N/A'\n",
    "#         print('MI: ', mi)\n",
    "#     else:\n",
    "#         mi = NMI(All[All[feature_name].notna()][feature_name].to_numpy().reshape(-1, 1), target.loc[All[All[feature_name].notna()].index])[0]\n",
    "#         print('MI: ', mi)\n",
    "\n",
    "#     MI_VAR.append((feature_name, notna, var, mi, 'Continues'))\n",
    "\n",
    "# df = pd.DataFrame(MI_VAR, columns=['feature_name', 'Not NA', 'var', 'MI', 'Category/Continues']).set_index('feature_name')\n",
    "for feature_name in feature_info.index:\n",
    "    if feature_name not in df.index:\n",
    "        df.loc[feature_name] = (All[feature_name].notna().sum(), None, None, None)\n",
    "        print(feature_name, '非NA数量:', All[feature_name].notna().sum())\n",
    "        \n",
    "df.to_csv(f'{task_name}/{task_name}_MI.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data2/huyaojun/.conda/envs/openmmlab/lib/python3.8/site-packages/sklearn/preprocessing/_discretization.py:239: FutureWarning: In version 1.5 onwards, subsample=200_000 will be used by default. Set subsample explicitly to silence this warning in the mean time. Set subsample=None to disable subsampling explicitly.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "required_header/download_date  非NA数量: 462610  NMI:  0.0\n",
      "required_header/link_text  非NA数量: 462610  NMI:  0.0\n",
      "sponsors/lead_sponsor/agency_class  非NA数量: 462610  NMI:  0.0054058517128195245\n",
      "overall_status  非NA数量: 462610  NMI:  0.007250214974096104\n",
      "phase  非NA数量: 184406  NMI:  0.0007327622719529863\n",
      "study_type  非NA数量: 462610  NMI:  0.014208919387675005\n",
      "has_expanded_access  非NA数量: 457832  NMI:  2.358518095060947e-05\n",
      "study_design_info/intervention_model  非NA数量: 354247  NMI:  0.0016817800587461025\n",
      "study_design_info/primary_purpose  非NA数量: 351664  NMI:  0.0005913787637894442\n",
      "eligibility/gender  非NA数量: 462607  NMI:  0.0001424368305320095\n",
      "eligibility/healthy_volunteers  非NA数量: 452269  NMI:  0.0005287985810707185\n",
      "study_first_posted{type}  非NA数量: 462610  NMI:  0.005350610425721624\n",
      "last_update_posted{type}  非NA数量: 462610  NMI:  0.0023829179289367153\n",
      "study_design_info/observational_model  非NA数量: 100987  NMI:  0.0028815387398530845\n",
      "study_design_info/time_perspective  非NA数量: 103164  NMI:  0.003916191833442467\n",
      "oversight_info/has_dmc  非NA数量: 385802  NMI:  0.00013776252975352215\n",
      "completion_date{type}  非NA数量: 453519  NMI:  0.005947338967043365\n",
      "primary_completion_date{type}  非NA数量: 456446  NMI:  0.006682971396105509\n",
      "enrollment{type}  非NA数量: 457024  NMI:  0.0034187034115119664\n",
      "biospec_retention  非NA数量: 22575  NMI:  0.016044720300179914\n",
      "eligibility/sampling_method  非NA数量: 103403  NMI:  0.0018669334681973688\n",
      "responsible_party/responsible_party_type  非NA数量: 431963  NMI:  0.00013738029419771615\n",
      "last_known_status  非NA数量: 60486  NMI:  0.006318464191098129\n",
      "study_design_info/allocation  非NA数量: 272488  NMI:  0.0022880919659943153\n",
      "start_date{type}  非NA数量: 292811  NMI:  0.004846433942162149\n",
      "results_first_posted{type}  非NA数量: 62022  NMI:  0.0008929819354041566\n",
      "clinical_results/baseline/analyzed_list/analyzed/scope  非NA数量: 62022  NMI:  5.468648286105498e-05\n",
      "clinical_results/reported_events/serious_events/default_assessment  非NA数量: 32160  NMI:  8.811186677713199e-05\n",
      "clinical_results/reported_events/other_events/default_assessment  非NA数量: 40976  NMI:  6.637292278984429e-05\n",
      "clinical_results/certain_agreements/pi_employee  非NA数量: 62022  NMI:  0.004231994102010433\n",
      "patient_data/sharing_ipd  非NA数量: 220603  NMI:  0.0003507239963519705\n",
      "oversight_info/is_fda_regulated_drug  非NA数量: 257973  NMI:  0.0005450737808320999\n",
      "oversight_info/is_fda_regulated_device  非NA数量: 257939  NMI:  0.0005244054249576031\n",
      "oversight_info/is_us_export  非NA数量: 37921  NMI:  3.5717736301438674e-05\n",
      "disposition_first_posted{type}  非NA数量: 8769  NMI:  0.000662371939031906\n",
      "eligibility/gender_based  非NA数量: 12750  NMI:  0.0\n",
      "expanded_access_info/expanded_access_type_intermediate  非NA数量: 0  NMI:  1.0\n",
      "expanded_access_info/expanded_access_type_treatment  非NA数量: 0  NMI:  1.0\n",
      "oversight_info/is_unapproved_device  非NA数量: 3437  NMI:  0.0\n",
      "expanded_access_info/expanded_access_type_individual  非NA数量: 1  NMI:  1.0\n",
      "oversight_info/is_ppsd  非NA数量: 12  NMI:  1.0\n"
     ]
    }
   ],
   "source": [
    "duration_target = pd.read_csv('trial-duration-prediction/output.csv', index_col=0)\n",
    "All = pd.concat(All_data)\n",
    "All = All.loc[duration_target.index]\n",
    "duration_target = duration_target['time_day']\n",
    "est = KBinsDiscretizer(n_bins=10, encode='ordinal', strategy='uniform')\n",
    "continuous_duration_discrete = est.fit_transform(duration_target.values.reshape(-1, 1)).flatten()\n",
    "for feature_name in features_category:\n",
    "    print(feature_name, end='  ')\n",
    "    print('非NA数量:', All[feature_name].notna().sum(), end='  ')\n",
    "    print('NMI: ', NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes, continuous_duration_discrete[All[feature_name].notna()]))\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "phase\n",
      "非NA数量: 42786\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data2/huyaojun/.conda/envs/openmmlab/lib/python3.8/site-packages/sklearn/metrics/cluster/_supervised.py:66: UserWarning: Clustering metrics expects discrete values but received multiclass values for label, and continuous values for target\n",
      "  warnings.warn(msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "NMI:  0.11842588227318589\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAApgAAAGwCAYAAAANJhUPAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAABtl0lEQVR4nO3deVxU5f4H8M/IMgwMzBiogDGIAuICXkjFJddQR5aybuRVEqibXhXhumDlUqEl2nVvcc0LhpUtbrj8JKDAJDUDTUwjXHJMUVwZEEEYzu8Pf5yfE4qIRwbh83695vXqnOeZc77PRPDpOc85IxMEQQARERERkURamLoAIiIiImpaGDCJiIiISFIMmEREREQkKQZMIiIiIpIUAyYRERERSYoBk4iIiIgkxYBJRERERJIyN3UB1LRVVVXh/PnzsLW1hUwmM3U5REREVAeCIKC4uBjOzs5o0eLB5yMZMOmROn/+PFxcXExdBhEREdXD2bNn8eSTTz7w+xgw6ZGytbUFcPsH1M7OzsTVEBERUV3o9Xq4uLiIf8cfFAMmPVLVl8Xt7OwYMImIiB4z9V3expt8iIiIiEhSDJhEREREJCleIqcG0X/2FzCTK0xdBhFRk5G9MNzUJRDdE2cwiYiIiEhSDJhEREREJCkGTCIiIiKSFAMmEREREUmKAZOIiIiIJMWASURERESSYsAkIiIiIkkxYBIRERGRpBgwiYiIiEhSDJhEREREJCkGTCIiIiKSFAMmEREREUmKAZOIiIiIJMWASURERESSYsAkIiIiIkkxYBIRERGRpBgwG4nExESo1WpTl0FERET00BgwG0hkZCRkMhlkMhksLS3h7u6OuXPnorKy0tSl1epf//oXOnToAIVCgVatWuG5557Db7/9ZuqyiIiIqBFjwGxAWq0WBQUFyM/Px7Rp0xAXF4eFCxeauqxaPfXUU0hISMDx48eRkpICQRAwdOhQGAwGU5dGREREjRQDZgOSy+VwdHSEq6srJkyYgICAACQnJxv1SUlJQadOnaBUKsVAWu3gwYMYMmQIHBwcoFKpMGDAAOTk5IjtgiAgLi4OGo0Gcrkczs7OiImJEdvLy8sRGxuLtm3bwsbGBv7+/sjIyKi15nHjxqF///5o164d/Pz88N577+Hs2bP4448/JPlMiIiIqOlhwDQhhUKBW7duidulpaVYtGgRkpKSsGfPHuh0OsTGxortxcXFiIiIwN69e7F//354eHggMDAQxcXFAIBNmzZh6dKlWL16NfLz87F161Z4e3uL7580aRL27duHjRs34siRIwgNDYVWq0V+fn6d6r1x4wYSEhLg5uYGFxeXu/YpLy+HXq83ehEREVHzYm7qApojQRCQnp6OlJQUREdHi/srKiqwatUqdOjQAcDtQDh37lyxffDgwUbHWbNmDdRqNTIzMxEcHAydTgdHR0cEBATAwsICGo0GPXv2BADodDokJCRAp9PB2dkZABAbG4vdu3cjISEB8fHx96x3xYoVeP3113Hjxg107NgRqampsLS0vGvf+fPnY86cOfX7YIiIiKhJ4AxmA9qxYweUSiWsrKwwfPhwjBw5EnFxcWK7tbW1GC4BwMnJCYWFheL2xYsXMXbsWHh4eEClUsHOzg4lJSXQ6XQAgNDQUNy8eRPt27fH2LFjsWXLFvEmotzcXBgMBnh6ekKpVIqvzMxMnDx5sta6w8LCcOjQIWRmZsLT0xMvvfQSysrK7tp3xowZKCoqEl9nz56t78dFREREjynOYDagQYMGYeXKlbC0tISzszPMzY0/fgsLC6NtmUwGQRDE7YiICFy5cgXLly+Hq6sr5HI5evfuLV5md3FxQV5eHtLS0pCamoqJEydi4cKFyMzMRElJCczMzJCdnQ0zMzOj8yiVylrrVqlUUKlU8PDwQK9evdCyZUts2bIFo0aNqtFXLpdDLpc/0OdCRERETQsDZgOysbGBu7t7vd+flZWFFStWIDAwEABw9uxZXL582aiPQqFASEgIQkJCEBUVBS8vL+Tm5sLX1xcGgwGFhYXo169fvWsQBAGCIKC8vLzexyAiIqKmjQHzMeLh4YGkpCR0794der0e06dPh0KhENsTExNhMBjg7+8Pa2trbNiwAQqFAq6urrC3t0dYWBjCw8OxePFi+Pr64tKlS0hPT4ePjw+CgoJqnO/UqVP48ssvMXToULRq1Qp//vknFixYAIVCIYZcIiIior/iGszHyLp163Dt2jX4+flhzJgxiImJQevWrcV2tVqNtWvXom/fvvDx8UFaWhq2b98Oe3t7AEBCQgLCw8Mxbdo0dOzYESNGjMDBgweh0Wjuej4rKyv88MMPCAwMhLu7O0aOHAlbW1v8+OOPRuclIiIiupNMuHORH5HE9Ho9VCoVukWvgplccf83EBFRnWQvDDd1CdSEVf/9Lioqgp2d3QO/nzOYRERERCQpBkwiIiIikhQDJhERERFJigGTiIiIiCTFgElEREREkmLAJCIiIiJJMWASERERkaQYMImIiIhIUgyYRERERCQpBkwiIiIikhQDJhERERFJigGTiIiIiCTFgElEREREkmLAJCIiIiJJmZu6AGoe9rw3CnZ2dqYug4iIiBoAZzCJiIiISFIMmEREREQkKQZMIiIiIpIUAyYRERERSYoBk4iIiIgkxYBJRERERJJiwCQiIiIiSTFgEhEREZGkGDCJiIiISFL8Jh9qEGcX9IKtlZlJzq15O9ck5yUiImquOINJRERERJJiwCQiIiIiSTFgEhEREZGkGDCJiIiISFIMmEREREQkKQZMIiIiIpIUAyYRERERSYoBk4iIiIgkxYBJRERERJJiwCQiIiIiSTFgEhEREZGkGDCJiIiISFIMmEREREQkKQZMIiIiIpIUAyYRERERSYoBk4iIiIgk1agDZmJiItRqtanLaBAZGRmQyWS4fv26qUshIiIieigmDZiRkZGQyWSQyWSwtLSEu7s75s6di8rKSlOWVaurV68iOjoaHTt2hEKhgEajQUxMDIqKimr0PXPmDBQKBUpKShAXFyeO1dzcHO3atcOUKVNQUlJiglHUXVxcHLy8vGBjY4OWLVsiICAABw4cMHVZRERE1IiZm7oArVaLhIQElJeXY9euXYiKioKFhQVmzJhh6tLu6vz58zh//jwWLVqEzp0748yZMxg/fjzOnz+Pb775xqjvtm3bMGjQICiVSgBAly5dkJaWhsrKSmRlZeHVV19FaWkpVq9ebYqh1Imnpyc++ugjtG/fHjdv3sTSpUsxdOhQnDhxAq1atTJ1eURERNQImfwSuVwuh6OjI1xdXTFhwgQEBAQgOTnZqE9KSgo6deoEpVIJrVaLgoICse3gwYMYMmQIHBwcoFKpMGDAAOTk5IjtgiAgLi4OGo0Gcrkczs7OiImJEdvLy8sRGxuLtm3bwsbGBv7+/sjIyLhnvV27dsWmTZsQEhKCDh06YPDgwZg3bx62b99eY+Z127ZtePbZZ8Vtc3NzODo64sknn8TIkSMRFhZWY6zZ2dno3r07rK2t0adPH+Tl5YltJ0+exHPPPYc2bdpAqVSiR48eSEtLM3r/ihUr4OHhASsrK7Rp0wYvvvii2FZVVYX58+fDzc0NCoUC3bp1qxGK/2r06NEICAhA+/bt0aVLFyxZsgR6vR5Hjhy5a//y8nLo9XqjFxERETUvJg+Yf6VQKHDr1i1xu7S0FIsWLUJSUhL27NkDnU6H2NhYsb24uBgRERHYu3cv9u/fDw8PDwQGBqK4uBgAsGnTJixduhSrV69Gfn4+tm7dCm9vb/H9kyZNwr59+7Bx40YcOXIEoaGh0Gq1yM/Pr3PNRUVFsLOzg7n5/08IX79+HXv37jUKmPcbKwDMmjULixcvxs8//wxzc3O8+uqrYltJSQkCAwORnp6OQ4cOQavVIiQkBDqdDgDw888/IyYmBnPnzkVeXh52796N/v37i++fP38+Pv30U6xatQq//vorpkyZgpdffhmZmZl1GuetW7ewZs0aqFQqdOvW7a595s+fD5VKJb5cXFzqdGwiIiJqOkx+ibyaIAhIT09HSkoKoqOjxf0VFRVYtWoVOnToAOB2IJw7d67YPnjwYKPjrFmzBmq1GpmZmQgODoZOp4OjoyMCAgJgYWEBjUaDnj17AgB0Oh0SEhKg0+ng7OwMAIiNjcXu3buRkJCA+Pj4+9Z9+fJlvPvuuxg3bpzR/l27dsHHx0c87l9lZ2fj888/r1H/vHnzMGDAAADAm2++iaCgIJSVlcHKygrdunUzCnbvvvsutmzZguTkZEyaNAk6nQ42NjYIDg6Gra0tXF1d4evrC+D2zGJ8fDzS0tLQu3dvAED79u2xd+9erF69Wjzn3ezYsQP/+Mc/UFpaCicnJ6SmpsLBweGufWfMmIGpU6eK23q9niGTiIiomTF5wNyxYweUSiUqKipQVVWF0aNHIy4uTmy3trYWwyUAODk5obCwUNy+ePEiZs+ejYyMDBQWFsJgMKC0tFSc1QsNDcWyZcvQvn17aLVaBAYGIiQkBObm5sjNzYXBYICnp6dRTeXl5bC3t79v7Xq9HkFBQejcubNRzUDNy+MAkJubC6VSCYPBgFu3biEoKAgfffSRUR8fHx+jsQJAYWEhNBqNeLPQzp07UVBQgMrKSty8eVMc65AhQ+Dq6iqOVavV4vnnn4e1tTVOnDiB0tJSDBkyxOh8t27dEkPovQwaNAiHDx/G5cuXsXbtWrz00ks4cOAAWrduXaOvXC6HXC6v/YMjIiKiJs3kAXPQoEFYuXIlLC0t4ezsbHSZGQAsLCyMtmUyGQRBELcjIiJw5coVLF++HK6urpDL5ejdu7d46dnFxQV5eXlIS0tDamoqJk6ciIULFyIzMxMlJSUwMzNDdnY2zMzMjM5TfWPOvRQXF0Or1cLW1hZbtmwxqvPWrVvYvXs3Zs6cafSejh07Ijk5Gebm5nB2doalpWWN4955HJlMBuD22kng9uxqamoqFi1aBHd3dygUCrz44oviWG1tbZGTk4OMjAx8++23ePvttxEXF4eDBw+Kd6vv3LkTbdu2NTrn/QKhjY0N3N3d4e7ujl69esHDwwPr1q1rtDdiERERkWmZPGBWh5f6ysrKwooVKxAYGAgAOHv2LC5fvmzUR6FQICQkBCEhIYiKioKXlxdyc3Ph6+sLg8GAwsJC9OvXr87n1Ov1GDZsGORyOZKTk2FlZWXUnpGRgZYtW9ZYp1j9KKb6ysrKQmRkJJ5//nkAt9dk/vHHH0Z9zM3NERAQgICAALzzzjtQq9X47rvvMGTIEMjlcuh0ulovh9dFVVUVysvLH+oYRERE1HSZPGA+LA8PDyQlJaF79+7Q6/WYPn06FAqF2J6YmAiDwQB/f39YW1tjw4YNUCgUcHV1hb29PcLCwhAeHo7FixfD19cXly5dQnp6Onx8fBAUFFTjfHq9HkOHDkVpaSk2bNhgdKd0q1atYGZmhuTk5Fpv7nmYsW7evBkhISGQyWR46623xNlN4PZyg1OnTqF///5o2bIldu3ahaqqKnTs2BG2traIjY3FlClTUFVVhaeffhpFRUXIysqCnZ0dIiIiapzvxo0bmDdvHp599lk4OTnh8uXL+Pjjj3Hu3DmEhoZKPj4iIiJqGh77gLlu3TqMGzcOfn5+cHFxQXx8vNFd5mq1GgsWLMDUqVNhMBjg7e2N7du3i2ssExIS8N5772HatGk4d+4cHBwc0KtXLwQHB9/1fDk5OeKDxv86G3n69Gm0a9cOycnJ+O9//yv5WJcsWYJXX30Vffr0gYODA9544w2jxwCp1Wps3rwZcXFxKCsrg4eHB7744gt06dIFwO2bglq1aoX58+fj1KlTUKvV8PPzq3Epv5qZmRl+++03rF+/HpcvX4a9vT169OiBH374QTwmERER0V/JhDsXNNJDy8nJweDBg3Hp0qUa60ebI71eD5VKhaMzOsHWyuz+b3gENG/nmuS8REREj6vqv9/Vj2J8UI3uOZiPu8rKSnz44YcMl0RERNRsPfaXyBubnj17is/ZJCIiImqOOINJRERERJJiwCQiIiIiSTFgEhEREZGkGDCJiIiISFIMmEREREQkKQZMIiIiIpIUAyYRERERSYoBk4iIiIgkxYBJRERERJJiwCQiIiIiSTFgEhEREZGk+F3k1CBc3twPOzs7U5dBREREDYAzmEREREQkKQZMIiIiIpIUAyYRERERSYoBk4iIiIgkxYBJRERERJJiwCQiIiIiSTFgEhEREZGkGDCJiIiISFIMmEREREQkKQZMIiIiIpIUvyqSGsSQVUNgruCPG1FdZUVnmboEIqJ64wwmEREREUmKAZOIiIiIJMWASURERESSYsAkIiIiIkkxYBIRERGRpBgwiYiIiEhSDJhEREREJCkGTCIiIiKSFAMmEREREUmKAZOIiIiIJMWASURERESSYsAkIiIiIkkxYBIRERGRpBgwiYiIiEhSDJhEREREJCkGTCIiIiKSFANmI5GYmAi1Wm3qMoiIiIgeGgNmA4mMjIRMJoNMJoOlpSXc3d0xd+5cVFZWmrq0OhEEAcOHD4dMJsPWrVtNXQ4RERE1YuamLqA50Wq1SEhIQHl5OXbt2oWoqChYWFhgxowZpi7tvpYtWwaZTGbqMoiIiOgxwBnMBiSXy+Ho6AhXV1dMmDABAQEBSE5ONuqTkpKCTp06QalUQqvVoqCgQGw7ePAghgwZAgcHB6hUKgwYMAA5OTliuyAIiIuLg0ajgVwuh7OzM2JiYsT28vJyxMbGom3btrCxsYG/vz8yMjLuW/fhw4exePFi/Pe//71v3/Lycuj1eqMXERERNS8MmCakUChw69Ytcbu0tBSLFi1CUlIS9uzZA51Oh9jYWLG9uLgYERER2Lt3L/bv3w8PDw8EBgaiuLgYALBp0yYsXboUq1evRn5+PrZu3Qpvb2/x/ZMmTcK+ffuwceNGHDlyBKGhodBqtcjPz79njaWlpRg9ejQ+/vhjODo63ndM8+fPh0qlEl8uLi71+WiIiIjoMcZL5CYgCALS09ORkpKC6OhocX9FRQVWrVqFDh06ALgdCOfOnSu2Dx482Og4a9asgVqtRmZmJoKDg6HT6eDo6IiAgABYWFhAo9GgZ8+eAACdToeEhATodDo4OzsDAGJjY7F7924kJCQgPj7+rrVOmTIFffr0wXPPPVensc2YMQNTp04Vt/V6PUMmERFRM8OA2YB27NgBpVKJiooKVFVVYfTo0YiLixPbra2txXAJAE5OTigsLBS3L168iNmzZyMjIwOFhYUwGAwoLS2FTqcDAISGhmLZsmVo3749tFotAgMDERISAnNzc+Tm5sJgMMDT09OopvLyctjb29+13uTkZHz33Xc4dOhQnccol8shl8vr3J+IiIiaHgbMBjRo0CCsXLkSlpaWcHZ2hrm58cdvYWFhtC2TySAIgrgdERGBK1euYPny5XB1dYVcLkfv3r3Fy+wuLi7Iy8tDWloaUlNTMXHiRCxcuBCZmZkoKSmBmZkZsrOzYWZmZnQepVJ513q/++47nDx5ssbjk/7+97+jX79+dVq/SURERM0PA2YDsrGxgbu7e73fn5WVhRUrViAwMBAAcPbsWVy+fNmoj0KhQEhICEJCQhAVFQUvLy/k5ubC19cXBoMBhYWF6NevX53O9+abb+K1114z2uft7Y2lS5ciJCSk3uMgIiKipo0B8zHi4eGBpKQkdO/eHXq9HtOnT4dCoRDbExMTYTAY4O/vD2tra2zYsAEKhQKurq6wt7dHWFgYwsPDsXjxYvj6+uLSpUtIT0+Hj48PgoKCapzP0dHxrjf2aDQauLm5PdKxEhER0eOLd5E/RtatW4dr167Bz88PY8aMQUxMDFq3bi22q9VqrF27Fn379oWPjw/S0tKwfft2cY1lQkICwsPDMW3aNHTs2BEjRozAwYMHodFoTDUkIiIiaoJkwp2L/IgkptfroVKp0PP9njBXcMKcqK6yorNMXQIRNWPVf7+LiopgZ2f3wO/nDCYRERERSYoBk4iIiIgkxYBJRERERJJiwCQiIiIiSTFgEhEREZGkGDCJiIiISFIMmEREREQkKQZMIiIiIpIUAyYRERERSYoBk4iIiIgkxYBJRERERJJiwCQiIiIiSTFgEhEREZGkzE1dADUPqeNTYWdnZ+oyiIiIqAFwBpOIiIiIJMWASURERESSYsAkIiIiIkkxYBIRERGRpBgwiYiIiEhSDJhEREREJCkGTCIiIiKSFAMmEREREUmKAZOIiIiIJMWASURERESS4ldFUoPYqx0OG3P+uBEREUllwJ5MU5dwT5zBJCIiIiJJMWASERERkaQYMImIiIhIUg8VMG/duoW8vDxUVlZKVQ8RERERPebqFTBLS0vxz3/+E9bW1ujSpQt0Oh0AIDo6GgsWLJC0QCIiIiJ6vNQrYM6YMQO//PILMjIyYGVlJe4PCAjAl19+KVlxRERERPT4qddzY7Zu3Yovv/wSvXr1gkwmE/d36dIFJ0+elKw4IiIiInr81GsG89KlS2jdunWN/Tdu3DAKnERERETU/NQrYHbv3h07d+4Ut6tD5SeffILevXtLUxkRERERPZbqdYk8Pj4ew4cPx7Fjx1BZWYnly5fj2LFj+PHHH5GZ2XifKk9EREREj169ZjCffvppHD58GJWVlfD29sa3336L1q1bY9++fXjqqaekrpGIiIiIHiP1/nLoDh06YO3atVLWQkRERERNQL1mMHNycpCbmytub9u2DSNGjMDMmTNx69YtyYojIiIiosdPvQLmv/71L/z+++8AgFOnTmHkyJGwtrbG119/jddff13SAomIiIjo8VKvgPn777/jb3/7GwDg66+/xoABA/D5558jMTERmzZtkrI+IiIiInrM1CtgCoKAqqoqAEBaWhoCAwMBAC4uLrh8+bJ01TUjiYmJUKvVpi6DiIiI6KHV+zmY7733HpKSkpCZmYmgoCAAwOnTp9GmTRtJC2wqIiMjIZPJIJPJYGlpCXd3d8ydOxeVlZWmLu2erl69iujoaHTs2BEKhQIajQYxMTEoKioydWlERETUiNXrLvJly5YhLCwMW7duxaxZs+Du7g4A+Oabb9CnTx9JC2xKtFotEhISUF5ejl27diEqKgoWFhaYMWOGqUu7q/Pnz+P8+fNYtGgROnfujDNnzmD8+PE4f/48vvnmG1OXR0RERI1UvWYwfXx8kJubi6KiIrzzzjvi/oULF2L9+vWSFdfUyOVyODo6wtXVFRMmTEBAQACSk5ON+qSkpKBTp05QKpXQarUoKCgQ2w4ePIghQ4bAwcEBKpUKAwYMQE5OjtguCALi4uKg0Wggl8vh7OyMmJgYsb28vByxsbFo27YtbGxs4O/vj4yMjHvW27VrV2zatAkhISHo0KEDBg8ejHnz5mH79u2NeuaViIiITKteAfNerKysYGFhIeUhmzSFQmH0WKfS0lIsWrQISUlJ2LNnD3Q6HWJjY8X24uJiREREYO/evdi/fz88PDwQGBiI4uJiAMCmTZuwdOlSrF69Gvn5+di6dSu8vb3F90+aNAn79u3Dxo0bceTIEYSGhkKr1SI/P7/ONRcVFcHOzg7m5nef/C4vL4derzd6ERERUfNSr0vkBoMBS5cuxVdffQWdTlfj2ZdXr16VpLimShAEpKenIyUlBdHR0eL+iooKrFq1Ch06dABwOxDOnTtXbB88eLDRcdasWQO1Wo3MzEwEBwdDp9PB0dERAQEBsLCwgEajQc+ePQEAOp0OCQkJ0Ol0cHZ2BgDExsZi9+7dSEhIQHx8/H3rvnz5Mt59912MGzfunn3mz5+POXPm1P3DICIioianXjOYc+bMwZIlSzBy5EgUFRVh6tSpeOGFF9CiRQvExcVJXGLTsWPHDiiVSlhZWWH48OEYOXKk0edlbW0thksAcHJyQmFhobh98eJFjB07Fh4eHlCpVLCzs0NJSQl0Oh0AIDQ0FDdv3kT79u0xduxYbNmyRbyUnZubC4PBAE9PTyiVSvGVmZmJkydP3rd2vV6PoKAgdO7cudZ/xzNmzEBRUZH4Onv27AN+SkRERPS4q9cM5meffYa1a9ciKCgIcXFxGDVqFDp06AAfHx/s37/faN0f/b9BgwZh5cqVsLS0hLOzc43LzH9dXiCTySAIgrgdERGBK1euYPny5XB1dYVcLkfv3r3FGWQXFxfk5eUhLS0NqampmDhxIhYuXIjMzEyUlJTAzMwM2dnZMDMzMzqPUqmste7i4mJotVrY2tpiy5YttS6DkMvlkMvldfo8iIiIqGmqV8C8cOGCuLZPqVSKj60JDg7GW2+9JV11TYyNjY14x319ZGVlYcWKFeJzR8+ePVvjuaMKhQIhISEICQlBVFQUvLy8kJubC19fXxgMBhQWFqJfv351Pqder8ewYcMgl8uRnJwMKyuretdPREREzUO9AuaTTz6JgoICaDQadOjQAd9++y38/Pxw8OBBzl49Qh4eHkhKSkL37t2h1+sxffp0KBQKsT0xMREGgwH+/v6wtrbGhg0boFAo4OrqCnt7e4SFhSE8PByLFy+Gr68vLl26hPT0dPj4+IjPMr2TXq/H0KFDUVpaig0bNhjdtNOqVasaM6FEREREQD0D5vPPP4/09HT4+/sjOjoaL7/8MtatWwedTocpU6ZIXSP9n3Xr1mHcuHHw8/ODi4sL4uPjje4yV6vVWLBgAaZOnQqDwQBvb29s374d9vb2AICEhAS89957mDZtGs6dOwcHBwf06tULwcHBdz1fTk4ODhw4AAA1Zl5Pnz6Ndu3aPZqBEhE1B3I5ZHZ2gExm6kroMVVWVlbv91pYWDzSiSKZcOciv3rat28f9u3bBw8PD4SEhEhRFzURer0eKpUKO3v3gc09Hm1ERNSsyGSwCHgGVv7+kPHRfvQQrBwdH+r9arUajo6OkN3lf3Kq/35XP57wQUnyF793797o3bu3FIciIiJq0iwCnoH1wIFopVZDbmYGzl9Sfdm4udXrfYIgoLS0VHxSjZOTk5RlAXiIgJmfn4/vv/8ehYWFqKqqMmp7++23H7owIiKiJkcuh5W/P1qp1bDj7CU9pIe58bb6Ho7CwkK0bt1a8svl9QqYa9euxYQJE+Dg4FBjalUmkzFgEhER3YXMzg4yCwvIeZMkNQLW1tYAbn/RS6MImO+99x7mzZuHN954Q9JiiIiImrT/m5DhZXFqDO629lIq9fomn2vXriE0NFTqWoiIiIioCahXwAwNDcW3334rdS1ERETUjHV95hl8vH69qcsgCdT5EvkHH3wg/rO7uzveeust7N+/H97e3jW+OpBfFUlERETUfNU5YC5dutRoW6lUIjMzE5mZmUb7ZTIZAyYRERFRM1bngHn69Om77q9+TvujXChKREREj7/A8HB08vAAAHyZnAxzc3P88x//wOyYGDFH3Cwrw8RZs7B1926oVSpMHz8er7z0kniMtxctwva0NJy/eBFtHBwQGhyMNydOFK+m5v72G96cPx+Hjh6FTCZDB1dXLJszB35duwIA9mVnI27pUhw6ehT2LVsiOCAAcVOmwOb/7qgmadRrDSZw+2sLu3btCisrK1hZWaFr16745JNPpKyNiIiImpgvtm6FuZkZvv/qK7w/cyY+Xr8e67/+Wmz/MCEBvl264IfNm/HaqFGYMmcO8u+Y5FLa2GDV/Pn4accOvD9zJtZ/843Rus3Xpk+Hc5s2yPj6a2R+8w2mjB0Li//7JrlTOh1eGDcOzw4Zgh+3bkXikiXYn52N2Pfea7gPoJmo12OK3n77bSxZsgTR0dHiN/js27cPU6ZMgU6nw9y5cyUtkoiIiJqGtk5OWDBjBmQyGTzc3HDs99/x8aefIvL/ZimH9u+PsaNHAwCmvPYaPl6/HnsOHIDH/31rzesTJojHcm3bFjGvvIJNu3Zh8muvAQD+LChAzD//Cc/27QEA7u3aif2XrFmDl4KDERURIbb9Z9YsDA8Px9J33oGVXP7Ix99c1Ctgrly5EmvXrsWoUaPEfc8++yx8fHwQHR3NgElERER31aNbN6NldT3/9jd8mJgIg8EAAOjasaPYJpPJ0MbBAZeuXBH3bdq1C6s2bMDps2dxo7QUlZWVsFUqxfaoyEhEv/UWvty2DQP79MGIYcPQXqMBAOTm5eHXvDx8tWOH2F8QBFRVVeHMn3+iY4cOj2zczU29AmZFRQW6d+9eY/9TTz2FysrKhy6KiIiImidzc+NoIpPJUPV/93scOHQIr73+OmZOmoRnnn4adkolNu3ahY8SE8X+MydNQmhQEFIyM5H6ww+I//BDJCxejJAhQ3CjtBSvjByJ8S+/XOO8Lo/g+7ibs3oFzDFjxmDlypVYsmSJ0f41a9YgLCxMksKIiIio6fn5l1+Mtg/+8gs6uLrW6asKfzp8GC7Ozpg+fry47+z58zX6ebi5wcPNDZMiI/HKtGnYsGULQoYMQbfOnZF34gQ6uLo+/ECoVvUKmMDtm3y+/fZb9OrVCwBw4MAB6HQ6hIeHY+rUqWK/v4ZQap6e3v0/sLOzM3UZREQmVVZWhtOnT8PGzQ1WVlamLqfBmVlb48/jx/HOmjX417/+hZycHKz+/HMsXrwYtl5ekFlYwKpNG9h6eYnvaSGXQ+7gAFsvL3Tt2xd/LlmCnYcPo0ePHti5cyd2fP890KIFbL28cPPmTUyfPh0vvvgi3Nzc8Oeff+JwXh7+/ve/w9bLC7PnzUOvXr0w46OP8Nprr8HGxgbHjh1DamoqPvroIxN+Mk1PvQLm0aNH4efnBwA4efIkAMDBwQEODg44evSo2I+PLiIiIqI7hYeH4+bNm+jZsyfMzMzw73//G+PGjavTe5999llMmTIFkyZNQnl5OYKCgvDWW28hLi4OAGBmZoYrV64gPDwcFy9ehIODA1544QXMmTMHAODj44PMzEzMmjUL/fr1gyAI6NChA0aOHPmohttsyYTqB1kSPQJ6vR4qlQpFRUWcwSSiZq96BtOtmc5gDhw4EH/729+wbNkyU5dCqP3n8WH/ftf7OZhERERERHfDgElEREREkqr3TT5EREREDyIjI8PUJVAD4QwmEREREUmKAZOIiIiIJMWASURERESSYsAkIiIiIkkxYBIRERGRpHgXOTWI1TP/Bwq5tanLoAY0aXGIqUsgIiIT4QwmERERPbTExESo1WpTl0GNBGcwiYiIGoGnpn/aoOfLXhj+QP0jIyOxfv16AICFhQU0Gg3Cw8Mxc+ZMmJs33jjxr3/9C2lpaTh//jyUSiX69OmD999/H15eXqYurUnjDCYRERHViVarRUFBAfLz8zFt2jTExcVh4cKFpi6rVk899RQSEhJw/PhxpKSkQBAEDB06FAaDwdSlNWkMmERERFQncrkcjo6OcHV1xYQJExAQEIDk5GSjPikpKejUqROUSqUYSKsdPHgQQ4YMgYODA1QqFQYMGICcnByxXRAExMXFQaPRQC6Xw9nZGTExMWJ7eXk5YmNj0bZtW9jY2MDf3/++3w40btw49O/fH+3atYOfnx/ee+89nD17Fn/88YcknwndHQMmERER1YtCocCtW7fE7dLSUixatAhJSUnYs2cPdDodYmNjxfbi4mJERERg79692L9/Pzw8PBAYGIji4mIAwKZNm7B06VKsXr0a+fn52Lp1K7y9vcX3T5o0Cfv27cPGjRtx5MgRhIaGQqvVIj8/v0713rhxAwkJCXBzc4OLi4tEnwLdTeNdNEFERESNkiAISE9PR0pKCqKjo8X9FRUVWLVqFTp06ADgdiCcO3eu2D548GCj46xZswZqtRqZmZkIDg6GTqeDo6MjAgICxHWePXv2BADodDokJCRAp9PB2dkZABAbG4vdu3cjISEB8fHx96x3xYoVeP3113Hjxg107NgRqampsLS0lOzzoJo4g0lERER1smPHDiiVSlhZWWH48OEYOXIk4uLixHZra2sxXAKAk5MTCgsLxe2LFy9i7Nix8PDwgEqlgp2dHUpKSqDT6QAAoaGhuHnzJtq3b4+xY8diy5YtqKysBADk5ubCYDDA09MTSqVSfGVmZuLkyZO11h0WFoZDhw4hMzMTnp6eeOmll1BWVibhJ0N/xRlMIiIiqpNBgwZh5cqVsLS0hLOzc427xy0sLIy2ZTIZBEEQtyMiInDlyhUsX74crq6ukMvl6N27t3iZ3cXFBXl5eUhLS0NqaiomTpyIhQsXIjMzEyUlJTAzM0N2djbMzMyMzqNUKmutW6VSQaVSwcPDA7169ULLli2xZcsWjBo16mE+DqoFAyYRERHViY2NDdzd3ev9/qysLKxYsQKBgYEAgLNnz+Ly5ctGfRQKBUJCQhASEoKoqCh4eXkhNzcXvr6+MBgMKCwsRL9+/epdgyAIEAQB5eXl9T4G3R8DJhERETUIDw8PJCUloXv37tDr9Zg+fToUCoXYnpiYCIPBAH9/f1hbW2PDhg1QKBRwdXWFvb09wsLCEB4ejsWLF8PX1xeXLl1Ceno6fHx8EBQUVON8p06dwpdffomhQ4eiVatW+PPPP7FgwQIoFAox5NKjwTWYRERE1CDWrVuHa9euwc/PD2PGjEFMTAxat24ttqvVaqxduxZ9+/aFj48P0tLSsH37dtjb2wMAEhISEB4ejmnTpqFjx44YMWIEDh48CI1Gc9fzWVlZ4YcffkBgYCDc3d0xcuRI2Nra4scffzQ6L0lPJty5OIJIYnq9HiqVCv+J2sjvIm9m+F3kRDWVlZXh9OnTcHNzg5WVlanLoWautp/H6r/fRUVFsLOze+BjcwaTiIiIiCTFgElEREREkmLAJCIiIiJJMWA2EomJiVCr1aYug4iIiOihMWA2kMjISMhkMshkMlhaWsLd3R1z584Vv6GgsVqzZg0GDhwIOzs7yGQyXL9+3dQlERERUSPHgNmAtFotCgoKkJ+fj2nTpiEuLg4LFy40dVm1Ki0thVarxcyZM01dChERET0mGDAbkFwuh6OjI1xdXTFhwgQEBAQgOTnZqE9KSgo6deoEpVIpBtJqBw8exJAhQ+Dg4ACVSoUBAwYgJydHbBcEAXFxcdBoNJDL5XB2dkZMTIzYXl5ejtjYWLRt2xY2Njbw9/dHRkZGrTVPnjwZb775Jnr16iXNh0BERERNHgOmCSkUCvH7V4Hbs4WLFi1CUlIS9uzZA51Oh9jYWLG9uLgYERER2Lt3L/bv3w8PDw8EBgaiuLgYALBp0yYsXboUq1evRn5+PrZu3Qpvb2/x/ZMmTcK+ffuwceNGHDlyBKGhodBqtcjPz5dsTOXl5dDr9UYvIiIial74VZEmIAgC0tPTkZKSgujoaHF/RUUFVq1ahQ4dOgC4HQjnzp0rtg8ePNjoOGvWrIFarUZmZiaCg4Oh0+ng6OiIgIAAWFhYQKPRoGfPngAAnU6HhIQE6HQ6ODs7AwBiY2Oxe/duJCQkID4+XpKxzZ8/H3PmzJHkWERERPR44gxmA9qxYweUSiWsrKwwfPhwjBw5EnFxcWK7tbW1GC4BwMnJCYWFheL2xYsXMXbsWHh4eEClUsHOzg4lJSXQ6XQAgNDQUNy8eRPt27fH2LFjsWXLFvEmotzcXBgMBnh6ekKpVIqvzMxMnDx5UrIxzpgxA0VFReLr7Nmzkh2biIiIHg+cwWxAgwYNwsqVK2FpaQlnZ2eYmxt//BYWFkbbMpkMd36TZ0REBK5cuYLly5fD1dUVcrkcvXv3Fi+zu7i4IC8vD2lpaUhNTcXEiROxcOFCZGZmoqSkBGZmZsjOzoaZmZnReZRKpWRjlMvlkMvlkh2PiKi50M31vn8nCWnezpX0eImJiZg8eXKzeNpIRkYGBg0ahGvXrvERg/fAGcwGZGNjA3d3d2g0mhrhsi6ysrIQExODwMBAdOnSBXK5HJcvXzbqo1AoEBISgg8++AAZGRnYt28fcnNz4evrC4PBgMLCQri7uxu9HB0dpRoiERE1UY/j4/auXr2K6OhodOzYEQqFAhqNBjExMSgqKqrR98yZM1AoFCgpKUFcXJw4VnNzc7Rr1w5TpkxBSUmJCUZRd3FxcfDy8oKNjQ1atmyJgIAAHDhwwCS1cAbzMeLh4YGkpCR0794der0e06dPh0KhENsTExNhMBjg7+8Pa2trbNiwAQqFAq6urrC3t0dYWBjCw8OxePFi+Pr64tKlS0hPT4ePjw+CgoLues4LFy7gwoULOHHiBIDbl9ptbW2h0WjwxBNPNMi4iYiocdBqtUhISEB5eTl27dqFqKgoWFhYYMaMGaYu7a7Onz+P8+fPY9GiRejcuTPOnDmD8ePH4/z58/jmm2+M+m7btg2DBg0Sr+p16dIFaWlpqKysRFZWFl599VWUlpZi9erVphhKnXh6euKjjz5C+/btcfPmTSxduhRDhw7FiRMn0KpVqwathTOYj5F169bh2rVr8PPzw5gxYxATE4PWrVuL7Wq1GmvXrkXfvn3h4+ODtLQ0bN++Hfb29gCAhIQEhIeHY9q0aejYsSNGjBiBgwcPQqPR3POcq1atgq+vL8aOHQsA6N+/P3x9fWs8XomIiJq+x+1xe127dsWmTZsQEhKCDh06YPDgwZg3bx62b99eY+Z127ZtePbZZ8Vtc3NzODo64sknn8TIkSMRFhZWY6zZ2dno3r07rK2t0adPH+Tl5YltJ0+exHPPPYc2bdpAqVSiR48eSEtLM3r/ihUr4OHhASsrK7Rp0wYvvvii2FZVVYX58+fDzc0NCoUC3bp1qxGK/2r06NEICAhA+/bt0aVLFyxZsgR6vR5Hjhyp9X2PAmcwG0hiYmKt7ZGRkYiMjDTaN2LECKM1mL6+vjh48KBRnzt/GEeMGIERI0bc8xwWFhaYM2fOA93lHRcXZ3QjEhERUTWFQoErV66I23c+bq9FixZ4+eWXERsbi88++wzA/z9u78MPP4QgCFi8eDECAwORn58PW1tb8XF7GzduRJcuXXDhwgX88ssv4vEnTZqEY8eOYePGjXB2dsaWLVug1WqRm5sLDw+POtVcVFQEOzs7o6Vq169fx969e5GUlFTrWO98tCAAzJo1C4sXL0arVq0wfvx4vPrqq8jKygIAlJSUIDAwEPPmzYNcLsenn36KkJAQ5OXlQaPR4Oeff0ZMTAySkpLQp08fXL16FT/88IN47Pnz52PDhg1YtWoVPDw8sGfPHrz88sto1aoVBgwYcN9x3rp1C2vWrIFKpUK3bt3q9NlIiQGTiIiIHsjj+ri9y5cv491338W4ceOM9u/atQs+Pj7icf8qOzsbn3/+eY36582bJ4a9N998E0FBQSgrK4OVlRW6detmFOzeffddbNmyBcnJyZg0aRJ0Oh1sbGwQHBwMW1tbuLq6wtfXF8Dtmdr4+HikpaWhd+/eAID27dtj7969WL16da0Bc8eOHfjHP/6B0tJSODk5ITU1FQ4ODvf9bKTGgElERER1Uv24vYqKClRVVWH06NEP/Li92bNnIyMjA4WFhTAYDCgtLTV63N6yZcvQvn17aLVaBAYGIiQkBObm5kaP27tTeXm5uBSsNnq9HkFBQejcuXONK3N/vTwO3L7nQKlUwmAw4NatWwgKCsJHH31k1MfHx8dorABQWFgIjUYj3iy0c+dOFBQUoLKyEjdv3hTHOmTIELi6uopj1Wq1eP7552FtbY0TJ06gtLQUQ4YMMTrfrVu3xBB6L4MGDcLhw4dx+fJlrF27Fi+99BIOHDhgtKSuITBgEhERUZ08ro/bKy4uhlarha2tLbZs2WJU561bt7B7927MnDnT6D0dO3ZEcnIyzM3N4ezsDEtLyxrHvfM4MpkMwO21k8Dt2dXU1FQsWrQI7u7uUCgUePHFF8Wx2traIicnBxkZGfj222/x9ttvIy4uDgcPHhTvVt+5cyfatm1rdM77PQqw+ok17u7u6NWrFzw8PLBu3boGvxGLAZOIiIjqpDq81FdWVhZWrFiBwMBAAMDZs2fv+bi9kJAQREVFwcvLq8bj9vr161fnc+r1egwbNgxyuRzJycmwsrIyas/IyEDLli1rrFOsfhRTfWVlZSEyMhLPP/88gNtrMv/44w+jPubm5ggICEBAQADeeecdqNVqfPfddxgyZAjkcjl0Ol2d1lvWpqqqCuXl5Q91jPpgwCQiIqIG0dCP29Pr9Rg6dChKS0uxYcMG6PV66PV6AECrVq1gZmaG5OTkGpfHpRrr5s2bERISAplMhrfeekuc3QRuLzc4deoU+vfvj5YtW2LXrl2oqqpCx44dYWtri9jYWEyZMgVVVVV4+umnUVRUhKysLNjZ2SEiIqLG+W7cuIF58+bh2WefhZOTEy5fvoyPP/4Y586dQ2hoqOTjux8GTCIiokZA6m/WaYzWrVuHcePGwc/PDy4uLoiPj0dsbKzYrlarsWDBAkydOhUGgwHe3t41Hrf33nvvYdq0aTh37hwcHBzQq1cvBAcH3/V8OTk54oPG/zobefr0abRr1w7Jycn473//K/lYlyxZgldffRV9+vSBg4MD3njjDTHcVo918+bNiIuLQ1lZGTw8PPDFF1+gS5cuAG7fFNSqVSvMnz8fp06dglqthp+fX41L+dXMzMzw22+/Yf369bh8+TLs7e3Ro0cP/PDDD+IxG5JMuHNxBJHE9Ho9VCoV/hO1EQq5tanLoQY0aXGIqUsganTKyspw+vRpuLm51bhUSw0vJycHgwcPxqVLl2qsH20Oavt5rP77Xf1YpwfFB60TERFRs1RZWYkPP/ywWYbLR42XyImIiKhZ6tmzp/icTZIWZzCJiIiISFIMmEREREQkKQZMIiIiIpIUAyYRERERSYoBk4iIiIgkxbvIqUH8K354vZ6jRURERI8fzmASERHRQ0tMTIRarTZ1GdRIcAaTiIioEej7Yd8GPV9WdNYD9Y+MjMT69esBABYWFtBoNAgPD8fMmTNhbt7444QgCAgMDMTu3buxZcsWjBgxwtQlNWmN/yeCiIiIGgWtVouEhASUl5dj165diIqKgoWFBWbMmGHq0u5r2bJlkMlkpi6j2eAlciIiIqoTuVwOR0dHuLq6YsKECQgICEBycrJRn5SUFHTq1AlKpRJarRYFBQVi28GDBzFkyBA4ODhApVJhwIAByMnJEdsFQUBcXBw0Gg3kcjmcnZ0RExMjtpeXlyM2NhZt27aFjY0N/P39kZGRcd+6Dx8+jMWLF+O///3vw38IVCcMmERERFQvCoUCt27dErdLS0uxaNEiJCUlYc+ePdDpdIiNjRXbi4uLERERgb1792L//v3w8PBAYGAgiouLAQCbNm3C0qVLsXr1auTn52Pr1q3w9vYW3z9p0iTs27cPGzduxJEjRxAaGgqtVov8/Px71lhaWorRo0fj448/hqOj4yP4FOhueImciIiIHoggCEhPT0dKSgqio6PF/RUVFVi1ahU6dOgA4HYgnDt3rtg+ePBgo+OsWbMGarUamZmZCA4Ohk6ng6OjIwICAsR1ntXfFa7T6ZCQkACdTgdnZ2cAQGxsLHbv3o2EhATEx8fftdYpU6agT58+eO655yT9DKh2DJhERERUJzt27IBSqURFRQWqqqowevRoxMXFie3W1tZiuAQAJycnFBYWitsXL17E7NmzkZGRgcLCQhgMBpSWlkKn0wEAQkNDsWzZMrRv3x5arRaBgYEICQmBubk5cnNzYTAY4OnpaVRTeXk57O3t71pvcnIyvvvuOxw6dEjCT4HqggGTiIiI6mTQoEFYuXIlLC0t4ezsXOPucQsLC6NtmUwGQRDE7YiICFy5cgXLly+Hq6sr5HI5evfuLV5md3FxQV5eHtLS0pCamoqJEydi4cKFyMzMRElJCczMzJCdnQ0zMzOj8yiVyrvW+9133+HkyZM1Hp/097//Hf369avT+k2qHwZMIiIiqhMbGxu4u7vX+/1ZWVlYsWIFAgMDAQBnz57F5cuXjfooFAqEhIQgJCQEUVFR8PLyQm5uLnx9fWEwGFBYWIh+/frV6XxvvvkmXnvtNaN93t7eWLp0KUJCQuo9Dro/BkxqEAvHjoHVX/7Plh6dWRu+MXUJREQ1eHh4ICkpCd27d4der8f06dOhUCjE9sTERBgMBvj7+8Pa2hobNmyAQqGAq6sr7O3tERYWhvDwcCxevBi+vr64dOkS0tPT4ePjg6CgoBrnc3R0vOuNPRqNBm5ubo90rM0dAyYREVEj8KAPPn8crVu3DuPGjYOfnx9cXFwQHx9vdJe5Wq3GggULMHXqVBgMBnh7e2P79u3iGsuEhAS89957mDZtGs6dOwcHBwf06tULwcHBphoS3YNMuHNxBJHE9Ho9VCoVZr/0LGcwGxBnMIkap7KyMpw+fRpubm6wsrIydTnUzNX281j997uoqAh2dnYPfGw+B5OIiIiIJMWASURERESSYsAkIiIiIkkxYBIRERGRpBgwiYiIiEhSDJhEREREJCkGTCIiIiKSFAMmEREREUmKAZOIiIiIJMWASURERA8tMTERarXa1GVQI8HvIiciImoEMvsPaNDzDdiT+UD9IyMjsX79egCAhYUFNBoNwsPDMXPmTJibN844cfXqVbzzzjv49ttvodPp0KpVK4wYMQLvvvsuVCqVqctr0hrnTwQRERE1OlqtFgkJCSgvL8euXbsQFRUFCwsLzJgxw9Sl3dX58+dx/vx5LFq0CJ07d8aZM2cwfvx4nD9/Ht98842py2vSeImciIiI6kQul8PR0RGurq6YMGECAgICkJycbNQnJSUFnTp1glKphFarRUFBgdh28OBBDBkyBA4ODlCpVBgwYABycnLEdkEQEBcXB41GA7lcDmdnZ8TExIjt5eXliI2NRdu2bWFjYwN/f39kZGTcs96uXbti06ZNCAkJQYcOHTB48GDMmzcP27dvR2VlpXQfDNXAgElERET1olAocOvWLXG7tLQUixYtQlJSEvbs2QOdTofY2Fixvbi4GBEREdi7dy/2798PDw8PBAYGori4GACwadMmLF26FKtXr0Z+fj62bt0Kb29v8f2TJk3Cvn37sHHjRhw5cgShoaHQarXIz8+vc81FRUWws7NrtJf1m4pGHTCb04LhP/74AzKZDIcPHzZ1KURERLUSBAFpaWlISUnB4MGDxf0VFRVYtWoVunfvDj8/P0yaNAnp6eli++DBg/Hyyy/Dy8sLnTp1wpo1a1BaWorMzNvrQXU6HRwdHREQEACNRoOePXti7NixYltCQgK+/vpr9OvXDx06dEBsbCyefvppJCQk1Knuy5cv491338W4ceMk/DTobkwaMCMjIyGTySCTyWBpaQl3d3fMnTu30U9br1mzBgMHDoSdnR1kMhmuX79+1343b96EjY0NTpw4gcTERHGsLVq0wJNPPolXXnkFhYWFDVv8A5o/fz569OgBW1tbtG7dGiNGjEBeXp6pyyIiIhPYsWMHlEolrKysMHz4cIwcORJxcXFiu7W1NTp06CBuOzk5Gf2du3jxIsaOHQsPDw+oVCrY2dmhpKQEOp0OABAaGoqbN2+iffv2GDt2LLZs2SJmgtzcXBgMBnh6ekKpVIqvzMxMnDx58r616/V6BAUFoXPnzkY106Nh8vnhx23BMHD7EoBWq4VWq621ztTUVLi6usLd3R179+6FnZ0d8vLyUFVVhV9++QWvvPIKzp8/j5SUlAas/sFkZmYiKioKPXr0QGVlJWbOnImhQ4fi2LFjsLGxMXV5RETUgAYNGoSVK1fC0tISzs7ONS4zW1hYGG3LZDIIgiBuR0RE4MqVK1i+fDlcXV0hl8vRu3dv8TK7i4sL8vLykJaWhtTUVEycOBELFy5EZmYmSkpKYGZmhuzsbJiZmRmdR6lU1lp3cXExtFotbG1tsWXLlhp1kvRMfon8cVswDACTJ0/Gm2++iV69etXab9u2bXj22WfFbZlMBkdHRzg7O2P48OGIiYlBWloabt68KfY5deoUBg0aBGtra3Tr1g379u0T265cuYJRo0ahbdu2sLa2hre3N7744gujc37zzTfw9vaGQqGAvb09AgICcOPGDbH9k08+QadOnWBlZQUvLy+sWLGi1jHs3r0bkZGR6NKlC7p164bExETodDpkZ2fX+j4iImp6bGxs4O7uDo1GU681jFlZWYiJiUFgYCC6dOkCuVyOy5cvG/VRKBQICQnBBx98gIyMDOzbtw+5ubnw9fWFwWBAYWEh3N3djV6Ojo73PKder8fQoUNhaWmJ5ORkWFlZPXDd9OBMHjD/6nFcMHw3VVVV2LFjB5577rlax1pVVWW0JGDWrFmIjY3F4cOH4enpiVGjRontZWVleOqpp7Bz504cPXoU48aNw5gxY/DTTz8BAAoKCjBq1Ci8+uqrOH78ODIyMvDCCy+I//f42Wef4e2338a8efNw/PhxxMfH46233hKfa1YXRUVFAIAnnnjiru3l5eXQ6/VGLyIiIgDw8PBAUlISjh8/jgMHDiAsLAwKhUJsT0xMxLp163D06FGcOnUKGzZsgEKhgKurKzw9PREWFobw8HBs3rwZp0+fxk8//YT58+dj586ddz1fdbi8ceMG1q1bB71ejwsXLuDChQswGAwNNexmyeSXyKsJgoD09HSkpKQgOjpa3F+9YLh6TcekSZMwd+5csf3OxcXA7fWRarUamZmZCA4ONlowXP1g2J49ewL4/wXDOp0Ozs7OAIDY2Fjs3r0bCQkJiI+Pr/d49u/fDwDw9/e/a3t+fr64ENrW1hZXrlwRzx8UFAQAmDNnDrp06YITJ07Ay8sLbdu2NQrX0dHRSElJwVdffYWePXuioKAAlZWVeOGFF+Dq6goARmH6nXfeweLFi/HCCy8AANzc3HDs2DGsXr0aERER9x1TVVUVJk+ejL59+6Jr16537TN//nzMmTPnvsciIiJjD/rg88fRunXrMG7cOPj5+cHFxQXx8fFGf9fUajUWLFiAqVOnwmAwwNvbG9u3b4e9vT0AICEhAe+99x6mTZuGc+fOwcHBAb169UJwcPBdz5eTk4MDBw4AANzd3Y3aTp8+jXbt2j2agZLpA2b1guGKigpUVVVh9OjRD7xgePbs2cjIyEBhYSEMBgNKS0uNFgwvW7YM7du3h1arRWBgIEJCQmBubm60YPhO5eXl4g9zfW3btg3BwcFo0eL/J4mLioqgVCpRVVWFsrIyPP300/jkk0+M3ufj42M0VgAoLCyEl5cXDAYD4uPj8dVXX+HcuXO4desWysvLYW1tDQDo1q0bnnnmGXh7e2PYsGEYOnQoXnzxRbRs2RI3btzAyZMn8c9//lO8Iw8AKisr6/xtBlFRUTh69Cj27t17zz4zZszA1KlTxW29Xg8XF5c6HZ+IiBqvxMTEWtsjIyMRGRlptG/EiBFGazB9fX1x8OBBoz4vvviiUf8RI0bc8xwWFhaYM2dOnScyBg4caHR+ajgmD5iP64Lh+0lOTsaCBQuM9tna2iInJwctWrSAk5OT0WWBu41XJpMBuD1zCAALFy7E8uXLsWzZMnh7e8PGxgaTJ08Wx2pmZobU1FT8+OOP+Pbbb/Hhhx9i1qxZOHDggBhC165dW2NW9a9jv5tJkyZhx44d2LNnD5588sl79pPL5ZDL5fc9HhERETVdJg+Y1QuG6ysrKwsrVqxAYGAgAODs2bP3XDAcEhKCqKgoeHl51Vgw3K9fv4cax53y8/Nx5swZDBkyxGh/ixYtHnqszz33HF5++WUAt4Pn77//js6dO4t9ZDIZ+vbti759++Ltt9+Gq6srtmzZgqlTp8LZ2RmnTp1CWFhYnc8pCAKio6OxZcsWZGRkwM3Nrd71ExERUfNg8oD5sKoXDHfv3h16vR7Tp0+vsWDYYDDA398f1tbWRguG7e3txQXDixcvhq+vLy5duoT09HT4+PiIayH/qnqB8IkTJwDcfjaXra0tNBoNnnjiCWzbtg0BAQHirKGUY/3mm2/w448/omXLlliyZAkuXrwoBswDBw4gPT0dQ4cORevWrXHgwAFcunQJnTp1AnB7TWdMTAxUKhW0Wi3Ky8vx888/49q1a0aXte8UFRWFzz//HNu2bYOtrS0uXLgAAFCpVHedgSUiIiJqdHeRP6h169bh2rVr8PPzw5gxYxATE4PWrVuL7Wq1GmvXrkXfvn3h4+ODtLS0GguGw8PDMW3aNHTs2BEjRozAwYMHodFo7nnOVatWwdfXV1zL2L9/f/j6+oqPV/rr44mkMnv2bPj5+WHYsGEYOHAgHB0djdaq2NnZYc+ePQgMDISnpydmz56NxYsXY/jw4QCA1157DZ988gkSEhLg7e2NAQMGIDExsdZZyZUrV6KoqAgDBw6Ek5OT+Pryyy8lHx8RERE1DTKBq18ldfnyZTg5OeHPP/9EmzZtTF2Oyen1eqhUKsx+6VlY8cG2DWbWhm9MXQIR3UVZWRlOnz4NNzc3Po+RTK62n8fqv9/V393+oB77GczG5urVq1iyZAnDJRERETVbj/0azMbG09OzxmOPiIiIiJoTzmASERERkaQYMImIiIhIUgyYRERE9NASExOhVqtNXQY1ElyDSURE1Ah8NG17g55v0uKQB+ofGRmJ9evXA7j9rXMajQbh4eGYOXNmjW/ha0zWrFmDzz//HDk5OSguLsa1a9cYhBsAZzCJiIioTrRaLQoKCpCfn49p06YhLi4OCxcuNHVZtSotLYVWq8XMmTNNXUqzwoBJREREdSKXy+Ho6AhXV1dMmDABAQEB4peMVEtJSUGnTp2gVCrFQFrt4MGDGDJkCBwcHKBSqTBgwADk5OSI7YIgIC4uDhqNBnK5HM7OzoiJiRHby8vLERsbi7Zt28LGxgb+/v7IyMiotebJkyfjzTffRK9evaT5EKhOGDCJiIioXhQKBW7duiVul5aWYtGiRUhKSsKePXug0+kQGxsrthcXFyMiIgJ79+7F/v374eHhgcDAQBQXFwMANm3ahKVLl2L16tXIz8/H1q1b4e3tLb5/0qRJ2LdvHzZu3IgjR44gNDQUWq0W+fn5DTdoqpPGu2iCiIiIGiVBEJCeno6UlBRER0eL+ysqKrBq1Sp06NABwO1AOHfuXLF98ODBRsdZs2YN1Go1MjMzERwcDJ1OB0dHRwQEBIjrPHv27AkA0Ol0SEhIgE6ng7OzMwAgNjYWu3fvRkJCAuLj4x/1sOkBcAaTiIiI6mTHjh1QKpWwsrLC8OHDMXLkSMTFxYnt1tbWYrgEACcnJxQWForbFy9exNixY+Hh4QGVSgU7OzuUlJRAp9MBAEJDQ3Hz5k20b98eY8eOxZYtW1BZWQkAyM3NhcFggKenJ5RKpfjKzMzEyZMnG+YDoDrjDCYRERHVyaBBg7By5UpYWlrC2dm5xt3jFhYWRtsymQyCIIjbERERuHLlCpYvXw5XV1fI5XL07t1bvMzu4uKCvLw8pKWlITU1FRMnTsTChQuRmZmJkpISmJmZITs7G2ZmZkbnUSqVj2jEVF8MmERERFQnNjY2cHd3r/f7s7KysGLFCgQGBgIAzp49i8uXLxv1USgUCAkJQUhICKKiouDl5YXc3Fz4+vrCYDCgsLAQ/fr1e6hx0KPHgElEREQNwsPDA0lJSejevTv0ej2mT58OhUIhticmJsJgMMDf3x/W1tbYsGEDFAoFXF1dYW9vj7CwMISHh2Px4sXw9fXFpUuXkJ6eDh8fHwQFBd31nBcuXMCFCxdw4sQJALcvtdva2kKj0eCJJ55okHE3RwyY1CCmr02CnZ2dqcsgImq0HvTB54+jdevWYdy4cfDz84OLiwvi4+ON7jJXq9VYsGABpk6dCoPBAG9vb2zfvh329vYAgISEBLz33nuYNm0azp07BwcHB/Tq1QvBwcH3POeqVaswZ84ccbt///7isSIjIx/NQAky4c7FEUQS0+v1UKlUKCoqYsAkomavrKwMp0+fhpubG6ysrExdDjVztf08Puzfb95FTkRERESSYsAkIiIiIkkxYBIRERGRpBgwiYiIiEhSDJhEREREJCkGTCIiIiKSFAMmEREREUmKAZOIiIiIJMVv8qEGkbcwE0orm3u2d5o1uAGrISIiokeJM5hEREREJCnOYBIRETUC815+sUHPN2vDN5IeLzExEZMnT8b169clPW5j9Mcff8DNzQ2HDh3C3/72N1OX0yhxBpOIiIjuKzIyEjKZDDKZDJaWlnB3d8fcuXNRWVlp6tJqtWbNGgwcOBB2dnaQyWT3DMA3b96EjY0NTpw4gcTERHGsLVq0wJNPPolXXnkFhYWFDVv8A5o/fz569OgBW1tbtG7dGiNGjEBeXp5JamHAJCIiojrRarUoKChAfn4+pk2bhri4OCxcuNDUZdWqtLQUWq0WM2fOrLVfamoqXF1d4e7uDgCws7NDQUEB/vzzT6xduxb/8z//gzFjxjREyfWWmZmJqKgo7N+/H6mpqaioqMDQoUNx48aNBq+FAZOIiIjqRC6Xw9HREa6urpgwYQICAgKQnJxs1CclJQWdOnWCUqkUA2m1gwcPYsiQIXBwcIBKpcKAAQOQk5MjtguCgLi4OGg0Gsjlcjg7OyMmJkZsLy8vR2xsLNq2bQsbGxv4+/sjIyOj1ponT56MN998E7169aq137Zt2/Dss8+K2zKZDI6OjnB2dsbw4cMRExODtLQ03Lx5U+xz6tQpDBo0CNbW1ujWrRv27dsntl25cgWjRo1C27ZtYW1tDW9vb3zxxRdG5/zmm2/g7e0NhUIBe3t7BAQEGIXBTz75BJ06dYKVlRW8vLywYsWKWsewe/duREZGokuXLujWrRsSExOh0+mQnZ1d6/seBQZMIiIiqheFQoFbt26J26WlpVi0aBGSkpKwZ88e6HQ6xMbGiu3FxcWIiIjA3r17sX//fnh4eCAwMBDFxcUAgE2bNmHp0qVYvXo18vPzsXXrVnh7e4vvnzRpEvbt24eNGzfiyJEjCA0NhVarRX5+/kONo6qqCjt27MBzzz1X61irqqqMlgTMmjULsbGxOHz4MDw9PTFq1CixvaysDE899RR27tyJo0ePYty4cRgzZgx++uknAEBBQQFGjRqFV199FcePH0dGRgZeeOEFCIIAAPjss8/w9ttvY968eTh+/Dji4+Px1ltvYf369XUeV1FREQDgiSeeeODP5GHxJh8iIiJ6IIIgID09HSkpKYiOjhb3V1RUYNWqVejQoQOA24Fw7ty5YvvgwcaPpFuzZg3UajUyMzMRHBwMnU4HR0dHBAQEwMLCAhqNBj179gQA6HQ6JCQkQKfTwdnZGQAQGxuL3bt3IyEhAfHx8fUez/79+wEA/v7+d23Pz8/HqlWr0L17d9ja2uLKlSvi+YOCggAAc+bMQZcuXXDixAl4eXmhbdu2RuE6OjoaKSkp+Oqrr9CzZ08UFBSgsrISL7zwAlxdXQHAKEy/8847WLx4MV544QUAgJubG44dO4bVq1cjIiLivmOqqqrC5MmT0bdvX3Tt2rUen8rDYcAkIiKiOtmxYweUSiUqKipQVVWF0aNHIy4uTmy3trYWwyUAODk5Gd0Yc/HiRcyePRsZGRkoLCyEwWBAaWkpdDodACA0NBTLli1D+/btodVqERgYiJCQEJibmyM3NxcGgwGenp5GNZWXl8Pe3v6hxrVt2zYEBwejRYv/v7BbVFQEpVKJqqoqlJWV4emnn8Ynn3xi9D4fHx+jsQJAYWEhvLy8YDAYEB8fj6+++grnzp3DrVu3UF5eDmtrawBAt27d8Mwzz8Db2xvDhg3D0KFD8eKLL6Jly5a4ceMGTp48iX/+858YO3aseI7KykqoVKo6jSkqKgpHjx7F3r176/25PAwGTCIiIqqTQYMGYeXKlbC0tISzszPMzY1jhIWFhdG2TCYTL/kCQEREBK5cuYLly5fD1dUVcrkcvXv3Fi+zu7i4IC8vD2lpaUhNTcXEiROxcOFCZGZmoqSkBGZmZsjOzoaZmZnReZRK5UONKzk5GQsWLDDaZ2tri5ycHLRo0QJOTk5QKBQ13nfneGUyGYDbM4cAsHDhQixfvhzLli2Dt7c3bGxsMHnyZHGsZmZmSE1NxY8//ohvv/0WH374IWbNmoUDBw6IIXTt2rU1ZlX/Ova7mTRpEnbs2IE9e/bgySeffIBPQjoMmERERFQnNjY24l3W9ZGVlYUVK1YgMDAQAHD27FlcvnzZqI9CoUBISAhCQkIQFRUFLy8v5ObmwtfXFwaDAYWFhejXr99DjeNO+fn5OHPmDIYMGWK0v0WLFg891ueeew4vv/wygNvB8/fff0fnzp3FPjKZDH379kXfvn3x9ttvw9XVFVu2bMHUqVPh7OyMU6dOISwsrM7nFAQB0dHR2LJlCzIyMuDm5lbv+h8WAyYREVEjIPWDzxsjDw8PJCUloXv37tDr9Zg+fbrRzGBiYiIMBgP8/f1hbW2NDRs2QKFQwNXVFfb29ggLC0N4eDgWL14MX19fXLp0Cenp6fDx8RHXQv7VhQsXcOHCBZw4cQIAkJubC1tbW2g0GjzxxBPYtm0bAgICxFlDKcf6zTff4Mcff0TLli2xZMkSXLx4UQyYBw4cQHp6OoYOHYrWrVvjwIEDuHTpEjp16gTg9prOmJgYqFQqaLValJeX4+eff8a1a9cwderUu54zKioKn3/+ObZt2wZbW1tcuHABAKBSqe46A/so8S5yIiIiahDr1q3DtWvX4OfnhzFjxiAmJgatW7cW29VqNdauXYu+ffvCx8cHaWlp2L59u7jGMiEhAeHh4Zg2bRo6duyIESNG4ODBg9BoNPc856pVq+Dr6yuuZezfvz98fX3Fxyv99fFEUpk9ezb8/PwwbNgwDBw4EI6OjhgxYoTYbmdnhz179iAwMBCenp6YPXs2Fi9ejOHDhwMAXnvtNXzyySdISEiAt7c3BgwYgMTExFpnJVeuXImioiIMHDgQTk5O4uvLL7+UfHz3IxPuXBxBJDG9Xg+VSoWfZidDaWVzz36dZg2+ZxsRUVNRVlaG06dPw83NDVZWVqYup9m7fPkynJyc8Oeff6JNmzamLqfB1fbzWP33u6ioCHZ2dg98bM5gEhERUbN09epVLFmypFmGy0eNazCJiIioWfL09Kzx2COSBmcwiYiIiEhSDJh1EBkZabQw91EYOHAgJk+e/EjPQURERNQQmkzAjIyMhEwmq/HSarWmLg0AjGpSqVTo27cvvvvuO1OXVauysjJERkbC29sb5ubmjzxkExERUdPQZAImAGi1WhQUFBi9vvjii3ofz2AwiE/kl0JCQgIKCgqQlZUFBwcHBAcH49SpU5IdX2oGgwEKhQIxMTEICAgwdTlERET0mGhSAVMul8PR0dHo1bJlS7F9yZIl4tc1ubi4YOLEiSgpKRHbExMToVarkZycjM6dO0Mul4vfj1rt008/hb29PcrLy432jxgxAmPGjKm1PrVaDUdHR3Tt2hUrV67EzZs3kZqaKrZXVVXh9ddfxxNPPAFHR0ej73etS/1nzpxBSEgIWrZsCRsbG3Tp0gW7du0S248ePYrhw4dDqVSiTZs2GDNmTI1vULiTjY0NVq5cibFjx8LR0bHWsVUrLy+HXq83ehEREVHz0qQC5v20aNECH3zwAX799VesX78e3333HV5//XWjPqWlpXj//ffxySef4NdffzV6ACwAhIaGwmAwiA9oBW5/sf3OnTvx6quv1rmW6ifqV38nKQCsX78eNjY2OHDgAP7zn/9g7ty5RgH0fvVHRUWhvLwce/bsQW5uLt5//33x+1mvX7+OwYMHw9fXFz///DN2796Nixcv4qWXXqpzzXUxf/58qFQq8eXi4iLp8YmIiKjxa1IBc8eOHVAqlUav+Ph4sX3y5MkYNGgQ2rVrh8GDB+O9997DV199ZXSMiooKrFixAn369EHHjh1rfHWUQqHA6NGjkZCQIO7bsGEDNBoNBg4cWKc6S0tLMXv2bJiZmWHAgAHifh8fH7zzzjvw8PBAeHg4unfvjvT09DrXr9Pp0LdvX3h7e6N9+/YIDg5G//79AQAfffQRfH19ER8fDy8vL/j6+uK///0vvv/+e/z+++91qrsuZsyYgaKiIvF19uxZyY5NRERNE2+mbXqa1HMwBw0ahJUrVxrte+KJJ8R/TktLw/z58/Hbb79Br9ejsrISZWVlKC0tFYOkpaUlfHx8aj3P2LFj0aNHD5w7dw5t27ZFYmKieJNRbUaNGgUzMzPcvHkTrVq1wrp164zO9dfzOjk5obCwsM71x8TEYMKECfj2228REBCAv//97+Ixf/nlF3z//ffijOadTp48KdlzwORyOeRyuSTHIiJqTo7Pa9gbPx/0G9QiIyOxfv36GvuHDRuG3bt3S1VWvd35N9jOzg5du3bFu+++i8GDG+83xZWVlWH8+PHIzs7G8ePHERwcjK1bt5q6LEk0qRlMGxsbuLu7G72qA+Yff/yB4OBg+Pj4YNOmTcjOzsbHH38MwPgytUKhuG9Q9PX1Rbdu3fDpp58iOzsbv/76KyIjI+9b39KlS3H48GFcuHABFy5cQEREhFG7hYWF0bZMJhNvMqpL/a+99hpOnTqFMWPGIDc3F927d8eHH34IACgpKUFISAgOHz5s9MrPzxdnOYmIiGrDm2ml1ZRvpm1SAbM22dnZqKqqwuLFi9GrVy94enri/Pnz9T7ea6+9hsTERCQkJCAgIKBOaw0dHR3h7u6OVq1aPfD56lq/i4sLxo8fj82bN2PatGlYu3YtAMDPzw+//vor2rVrVyOE29jc+zvCiYiIqvFmWtPfTPu4aFIBs7y8XJwdrH5V/4t1d3dHRUUFPvzwQ5w6dQpJSUlYtWpVvc81evRo/Pnnn1i7du0D3dxTX3Wpf/LkyUhJScHp06eRk5OD77//Hp06dQJw+wagq1evYtSoUTh48CBOnjyJlJQUvPLKKzAYDPc877Fjx3D48GFcvXoVRUVF4swnERHRX/Fm2kd/M+3jokkFzN27d8PJycno9fTTTwMAunXrhiVLluD9999H165d8dlnn2H+/Pn1PpdKpcLf//53KJXKBnkAeV3qNxgMiIqKQqdOnaDVauHp6YkVK1YAAJydnZGVlQWDwYChQ4fC29sbkydPhlqtRosW9/4xCAwMhK+vL7Zv346MjAz4+vrC19f3kY6ViIgaJ95Ma/qbaR8XTeYmn8TERCQmJtbaZ8qUKZgyZYrRvjun2yMjI++6lvJexz137hzCwsLqdFOLIAi1tmdkZNTY99eFvverv3q95b14eHhg8+bNtRf6F3/88ccD9ScioqaLN9Oa/mbax0WTmsFsKNeuXcOWLVuQkZGBqKgoU5dDRETUIHgzLW+mrasmM4PZkHx9fXHt2jW8//776Nixo6nLISIiMrk7b0atXnr118vjD+K1117DsmXLcO7cuQe+mbY+6lp/9c2048ePx4wZM7B27VpER0fDz88PmzZtQrt27WBuznjFGcx6+OOPP1BUVITY2FhTl0JERNRgeDMtb6atK0ZsIiKiRuBBH3xuCtU3096pY8eO+O2334xuRp0xYwb69++P+fPnIzw8vF7nqr6ZdufOnQ1+M+296q++mfbPP/+EnZ0dtFotli5dCuD/b6Z94403MHToUJSXl8PV1RVarfa+N9OeOXNG3K6+kfZ+9240djLhcR8BNWp6vR4qlQo/zU6G0urez9t8HH6xEhE9rLKyMpw+fRpubm6wsrIydTmN3jPPPIMuXbrggw8+MHUpTVJtP4/Vf7+LiopgZ2f3wMfmDCYRERE1KteuXUNGRgYyMjLEx+3R44UBk4iIiBoV3kz7+GPAJCIiokaFz2B+/PEuciIiIiKSFAMmERFRA+P9tdQYPMqfQwZMIiKiBlL9TTKlpaUmroTo/38O//oNR1LgGkwiIqIGYmZmBrVaLX7/tbW19X2/NpFIaoIgoLS0FIWFhVCr1TAzM5P8HAyYREREDcjR0REAxJBJZCpqtVr8eZQaH7ROj9TDPqiViKipMhgMqKioMHUZ1ExZWFjUOnPJB60TERE9hszMzB7JpUmixoA3+RARERGRpBgwiYiIiEhSDJhEREREJCmuwaRHqvoeMr1eb+JKiIiIqK6q/27X915wBkx6pK5cuQIAcHFxMXElRERE9KCKi4uhUqke+H0MmPRIPfHEEwAAnU5Xrx/Qx41er4eLiwvOnj3bLB7LxPE2bRxv09WcxgpwvPUhCAKKi4vh7Oxcr/czYNIj1aLF7WW+KpWqWfxHXc3Ozo7jbcI43qatOY23OY0V4Hgf1MNMDPEmHyIiIiKSFAMmEREREUmKAZMeKblcjnfeeQdyudzUpTQIjrdp43ibtuY03uY0VoDjNQV+FzkRERERSYozmEREREQkKQZMIiIiIpIUAyYRERERSYoBk4iIiIgkxYBJj9THH3+Mdu3awcrKCv7+/vjpp59MXVKt5s+fjx49esDW1hatW7fGiBEjkJeXZ9SnrKwMUVFRsLe3h1KpxN///ndcvHjRqI9Op0NQUBCsra3RunVrTJ8+HZWVlUZ9MjIy4OfnB7lcDnd3dyQmJj7q4d3XggULIJPJMHnyZHFfUxvvuXPn8PLLL8Pe3h4KhQLe3t74+eefxXZBEPD222/DyckJCoUCAQEByM/PNzrG1atXERYWBjs7O6jVavzzn/9ESUmJUZ8jR46gX79+sLKygouLC/7zn/80yPjuZDAY8NZbb8HNzQ0KhQIdOnTAu+++a/Tdwo/zePfs2YOQkBA4OztDJpNh69atRu0NObavv/4aXl5esLKygre3N3bt2tWg462oqMAbb7wBb29v2NjYwNnZGeHh4Th//nyTHO9fjR8/HjKZDMuWLTPa39TGe/z4cTz77LNQqVSwsbFBjx49oNPpxPZG9ftaIHpENm7cKFhaWgr//e9/hV9//VUYO3asoFarhYsXL5q6tHsaNmyYkJCQIBw9elQ4fPiwEBgYKGg0GqGkpETsM378eMHFxUVIT08Xfv75Z6FXr15Cnz59xPbKykqha9euQkBAgHDo0CFh165dgoODgzBjxgyxz6lTpwRra2th6tSpwrFjx4QPP/xQMDMzE3bv3t2g473TTz/9JLRr107w8fER/v3vf4v7m9J4r169Kri6ugqRkZHCgQMHhFOnTgkpKSnCiRMnxD4LFiwQVCqVsHXrVuGXX34Rnn32WcHNzU24efOm2Eer1QrdunUT9u/fL/zwww+Cu7u7MGrUKLG9qKhIaNOmjRAWFiYcPXpU+OKLLwSFQiGsXr26Qcc7b948wd7eXtixY4dw+vRp4euvvxaUSqWwfPnyJjHeXbt2CbNmzRI2b94sABC2bNli1N5QY8vKyhLMzMyE//znP8KxY8eE2bNnCxYWFkJubm6Djff69etCQECA8OWXXwq//fabsG/fPqFnz57CU089ZXSMpjLeO23evFno1q2b4OzsLCxdurTJjvfEiRPCE088IUyfPl3IyckRTpw4IWzbts3ob2pj+n3NgEmPTM+ePYWoqChx22AwCM7OzsL8+fNNWNWDKSwsFAAImZmZgiDc/iVuYWEhfP3112Kf48ePCwCEffv2CYJw+5dEixYthAsXLoh9Vq5cKdjZ2Qnl5eWCIAjC66+/LnTp0sXoXCNHjhSGDRv2qId0V8XFxYKHh4eQmpoqDBgwQAyYTW28b7zxhvD000/fs72qqkpwdHQUFi5cKO67fv26IJfLhS+++EIQBEE4duyYAEA4ePCg2Od//ud/BJlMJpw7d04QBEFYsWKF0LJlS3H81efu2LGj1EOqVVBQkPDqq68a7XvhhReEsLAwQRCa1nj/+ge5Icf20ksvCUFBQUb1+Pv7C//6178kHeOdagtc1X766ScBgHDmzBlBEJrmeP/880+hbdu2wtGjRwVXV1ejgNnUxjty5Ejh5Zdfvud7Gtvva14ip0fi1q1byM7ORkBAgLivRYsWCAgIwL59+0xY2YMpKioCADzxxBMAgOzsbFRUVBiNy8vLCxqNRhzXvn374O3tjTZt2oh9hg0bBr1ej19//VXsc+cxqvuY6rOJiopCUFBQjZqa2niTk5PRvXt3hIaGonXr1vD19cXatWvF9tOnT+PChQtGtapUKvj7+xuNV61Wo3v37mKfgIAAtGjRAgcOHBD79O/fH5aWlmKfYcOGIS8vD9euXXvUwxT16dMH6enp+P333wEAv/zyC/bu3Yvhw4cDaHrjvVNDjq2x/Hz/VVFREWQyGdRqNYCmN96qqiqMGTMG06dPR5cuXWq0N6XxVlVVYefOnfD09MSwYcPQunVr+Pv7G11Gb2y/rxkw6ZG4fPkyDAaD0Q8xALRp0wYXLlwwUVUPpqqqCpMnT0bfvn3RtWtXAMCFCxdgaWkp/sKudue4Lly4cNdxV7fV1kev1+PmzZuPYjj3tHHjRuTk5GD+/Pk12praeE+dOoWVK1fCw8MDKSkpmDBhAmJiYrB+/Xqjemv7ub1w4QJat25t1G5ubo4nnnjigT6ThvDmm2/iH//4B7y8vGBhYQFfX19MnjwZYWFhRrU0lfHeqSHHdq8+pvxdV1ZWhjfeeAOjRo2CnZ0dgKY33vfffx/m5uaIiYm5a3tTGm9hYSFKSkqwYMECaLVafPvtt3j++efxwgsvIDMzU6yzMf2+Nn+gERI1I1FRUTh69Cj27t1r6lIembNnz+Lf//43UlNTYWVlZepyHrmqqip0794d8fHxAABfX18cPXoUq1atQkREhImrk95XX32Fzz77DJ9//jm6dOmCw4cPY/LkyXB2dm6S46XbKioq8NJLL0EQBKxcudLU5TwS2dnZWL58OXJyciCTyUxdziNXVVUFAHjuuecwZcoUAMDf/vY3/Pjjj1i1ahUGDBhgyvLuijOY9Eg4ODjAzMysxt1rFy9ehKOjo4mqqrtJkyZhx44d+P777/Hkk0+K+x0dHXHr1i1cv37dqP+d43J0dLzruKvbautjZ2cHhUIh9XDuKTs7G4WFhfDz84O5uTnMzc2RmZmJDz74AObm5mjTpk2TGq+TkxM6d+5stK9Tp07iXZjV9db2c+vo6IjCwkKj9srKSly9evWBPpOGMH36dHEW09vbG2PGjMGUKVPE2eqmNt47NeTY7tXHFGOvDpdnzpxBamqqOHsJNK3x/vDDDygsLIRGoxF/d505cwbTpk1Du3btxDqbyngdHBxgbm5+399fjen3NQMmPRKWlpZ46qmnkJ6eLu6rqqpCeno6evfubcLKaicIAiZNmoQtW7bgu+++g5ubm1H7U089BQsLC6Nx5eXlQafTiePq3bs3cnNzjX6xVf+ir/7l0Lt3b6NjVPdp6M/mmWeeQW5uLg4fPiy+unfvjrCwMPGfm9J4+/btW+OxU7///jtcXV0BAG5ubnB0dDSqVa/X48CBA0bjvX79OrKzs8U+3333HaqqquDv7y/22bNnDyoqKsQ+qamp6NixI1q2bPnIxvdXpaWlaNHC+Ne8mZmZOBvS1MZ7p4YcW2P5+a4Ol/n5+UhLS4O9vb1Re1Ma75gxY3DkyBGj313Ozs6YPn06UlJSxDqbyngtLS3Ro0ePWn9/Nbq/Tw90SxDRA9i4caMgl8uFxMRE4dixY8K4ceMEtVptdPdaYzNhwgRBpVIJGRkZQkFBgfgqLS0V+4wfP17QaDTCd999J/z8889C7969hd69e4vt1Y+BGDp0qHD48GFh9+7dQqtWre76GIjp06cLx48fFz7++GOTP6ao2p13kQtC0xrvTz/9JJibmwvz5s0T8vPzhc8++0ywtrYWNmzYIPZZsGCBoFarhW3btglHjhwRnnvuubs+2sbX11c4cOCAsHfvXsHDw8Po0SfXr18X2rRpI4wZM0Y4evSosHHjRsHa2rrBH1MUEREhtG3bVnxM0ebNmwUHBwfh9ddfbxLjLS4uFg4dOiQcOnRIACAsWbJEOHTokHjXdEONLSsrSzA3NxcWLVokHD9+XHjnnXceyWNsahvvrVu3hGeffVZ48sknhcOHDxv9/rrzDummMt67+etd5E1tvJs3bxYsLCyENWvWCPn5+eLjg3744QfxGI3p9zUDJj1SH374oaDRaARLS0uhZ8+ewv79+01dUq0A3PWVkJAg9rl586YwceJEoWXLloK1tbXw/PPPCwUFBUbH+eOPP4Thw4cLCoVCcHBwEKZNmyZUVFQY9fn++++Fv/3tb4KlpaXQvn17o3OY0l8DZlMb7/bt24WuXbsKcrlc8PLyEtasWWPUXlVVJbz11ltCmzZtBLlcLjzzzDNCXl6eUZ8rV64Io0aNEpRKpWBnZye88sorQnFxsVGfX375RXj66acFuVwutG3bVliwYMEjH9tf6fV64d///reg0WgEKysroX379sKsWbOMAsfjPN7vv//+rv+9RkRENPjYvvrqK8HT01OwtLQUunTpIuzcubNBx3v69Ol7/v76/vvvm9x47+ZuAbOpjXfdunWCu7u7YGVlJXTr1k3YunWr0TEa0+9rmSDc8ZUOREREREQPiWswiYiIiEhSDJhEREREJCkGTCIiIiKSFAMmEREREUmKAZOIiIiIJMWASURERESSYsAkIiIiIkkxYBIRERGRpBgwiYiIiEhSDJhERNTg/vjjD8hkMhw+fNjUpRDRI8CASURERESSYsAkImqGqqqq8J///Afu7u6Qy+XQaDSYN28eACA3NxeDBw+GQqGAvb09xo0bh5KSEvG9AwcOxOTJk42ON2LECERGRorb7dq1Q3x8PF599VXY2tpCo9FgzZo1YrubmxsAwNfXFzKZDAMHDnxkYyWihseASUTUDM2YMQMLFizAW2+9hWPHjuHzzz9HmzZtcOPGDQwbNgwtW7bEwYMH8fXXXyMtLQ2TJk164HMsXrwY3bt3x6FDhzBx4kRMmDABeXl5AICffvoJAJCWloaCggJs3rxZ0vERkWmZm7oAIiJqWMXFxVi+fDk++ugjREREAAA6dOiAp59+GmvXrkVZWRk+/fRT2NjYAAA++ugjhISE4P3330ebNm3qfJ7AwEBMnDgRAPDGG29g6dKl+P7779GxY0e0atUKAGBvbw9HR0eJR0hEpsYZTCKiZub48eMoLy/HM888c9e2bt26ieESAPr27Yuqqipx9rGufHx8xH+WyWRwdHREYWFh/QsnoscGAyYRUTOjUCge6v0tWrSAIAhG+yoqKmr0s7CwMNqWyWSoqqp6qHMT0eOBAZOIqJnx8PCAQqFAenp6jbZOnTrhl19+wY0bN8R9WVlZaNGiBTp27AgAaNWqFQoKCsR2g8GAo0ePPlANlpaW4nuJqOlhwCQiamasrKzwxhtv4PXXX8enn36KkydPYv/+/Vi3bh3CwsJgZWWFiIgIHD16FN9//z2io6MxZswYcf3l4MGDsXPnTuzcuRO//fYbJkyYgOvXrz9QDa1bt4ZCocDu3btx8eJFFBUVPYKREpGpMGASETVDb731FqZNm4a3334bnTp1wsiRI1FYWAhra2ukpKTg6tWr6NGjB1588UU888wz+Oijj8T3vvrqq4iIiEB4eDgGDBiA9u3bY9CgQQ90fnNzc3zwwQdYvXo1nJ2d8dxzz0k9RCIyIZnw14U0REREREQPgTOYRERERCQpBkwiIiIikhQDJhERERFJigGTiIiIiCTFgElEREREkmLAJCIiIiJJMWASERERkaQYMImIiIhIUgyYRERERCQpBkwiIiIikhQDJhERERFJ6n8BPskMUUoalxwAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "feature_name = 'phase'\n",
    "print(feature_name)\n",
    "print('非NA数量:', All[feature_name].notna().sum())\n",
    "print('NMI: ', NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes, continuous_target_discrete[All[feature_name].notna()]))\n",
    "sns.countplot(data=All, y=feature_name, hue=feature_name)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sponsors/lead_sponsor/agency_class\n",
      "非NA数量: 62058\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data2/huyaojun/.conda/envs/openmmlab/lib/python3.8/site-packages/sklearn/metrics/cluster/_supervised.py:66: UserWarning: Clustering metrics expects discrete values but received multiclass values for label, and continuous values for target\n",
      "  warnings.warn(msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "NMI:  0.06714869072644133\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "feature_name = 'sponsors/lead_sponsor/agency_class'\n",
    "print(feature_name)\n",
    "print('非NA数量:', All[feature_name].notna().sum())\n",
    "print('NMI: ', NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes, continuous_target_discrete[All[feature_name].notna()]))\n",
    "sns.countplot(data=All, y=feature_name, hue=feature_name)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "study_type\n",
      "非NA数量: 62058\n",
      "NMI:  0.002811026538073855\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "feature_name = 'study_type'\n",
    "print(feature_name)\n",
    "print('非NA数量:', All[feature_name].notna().sum())\n",
    "print('NMI: ', NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes, continuous_target_discrete[All[feature_name].notna()]))\n",
    "sns.countplot(data=All, y=feature_name, hue=feature_name)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "has_expanded_access\n",
      "非NA数量: 61574\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data2/huyaojun/.conda/envs/openmmlab/lib/python3.8/site-packages/sklearn/metrics/cluster/_supervised.py:66: UserWarning: Clustering metrics expects discrete values but received binary values for label, and continuous values for target\n",
      "  warnings.warn(msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "NMI:  0.006122887737182305\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "feature_name = 'has_expanded_access'\n",
    "print(feature_name)\n",
    "print('非NA数量:', All[feature_name].notna().sum())\n",
    "print('NMI: ', NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes, continuous_target_discrete[All[feature_name].notna()]))\n",
    "sns.countplot(data=All, y=feature_name, hue=feature_name)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "study_design_info/intervention_model\n",
      "非NA数量: 58441\n",
      "NMI:  0.16559458016578063\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "feature_name = 'study_design_info/intervention_model'\n",
    "print(feature_name)\n",
    "print('非NA数量:', All[feature_name].notna().sum())\n",
    "print('NMI: ', NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes, continuous_target_discrete[All[feature_name].notna()]))\n",
    "sns.countplot(data=All, y=feature_name, hue=feature_name)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "study_design_info/primary_purpose\n",
      "非NA数量: 57781\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data2/huyaojun/.conda/envs/openmmlab/lib/python3.8/site-packages/sklearn/metrics/cluster/_supervised.py:66: UserWarning: Clustering metrics expects discrete values but received multiclass values for label, and continuous values for target\n",
      "  warnings.warn(msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "NMI:  0.07574818661305258\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAtEAAAGwCAYAAABviTllAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAACsh0lEQVR4nOzde1yP9//48cdb6XxySiEpFWVF5FQjjFXMsWHWRqgZWkIWMzlMThNm5vBhK8awzWHG5tS0EZtj2GoOyWkyZqrlkNT1+8PP9fVWqLcoPO+32/u29/u6Xtfr9byurPez1/W6Xi+NoigKQgghhBBCiGKrUNYBCCGEEEII8ayRJFoIIYQQQogSkiRaCCGEEEKIEpIkWgghhBBCiBKSJFoIIYQQQogSkiRaCCGEEEKIEpIkWgghhBBCiBLSL+sAhHheFRQUcOHCBczNzdFoNGUdjhBCCCGKQVEU/vvvP2rUqEGFCg/ub5YkWogn5MKFC9jZ2ZV1GEIIIYTQwblz56hVq9YD90sSLcQTYm5uDtz5n9DCwqKMoxFCCCFEcWRnZ2NnZ6d+jz+IJNFCPCF3h3BYWFhIEi2EEEI8Yx41FFMeLBRCCCGEEKKEpCdaiCes9Ycr0TM0LuswnogDH/ct6xCEEEKIMiE90UIIIYQQQpSQJNFCCCGEEEKUkCTRQgghhBBClJAk0UIIIYQQQpSQJNFCCCGEEEKUkCTRQgghhBBClJAk0UIIIYQQQpSQJNFCCCGEEEKUkCTRQgghhBBClJAk0UIIIYQQQpSQJNFCCCGEEEKUkCTR4rkXHBxMt27dyjoMIYQQQjxHJIkWhQQHB6PRaNBoNBgYGODk5MSkSZO4fft2WYf2UKdPn0aj0ZCcnKy1/ZNPPiE+Pr5MYhJCCCHE80m/rAMQ5ZO/vz9xcXHk5ubyww8/MHToUCpWrMiYMWO0yt26dQsDA4MyirJ4LC0tyzoEIYQQQjxnpCdaFMnQ0BAbGxvs7e0ZPHgw7du3Z8OGDerQiJiYGGrUqEG9evUAOHfuHL169cLKyorKlSvTtWtXTp8+DcDWrVsxMjIiMzNTq41hw4bRrl079fOuXbto1aoVxsbG2NnZER4ezrVr19T9derUYcqUKQwYMABzc3Nq167N//73P3W/g4MDAJ6enmg0Gtq0aQMUHs6Rm5tLeHg41tbWGBkZ8fLLL7Nv3z51f2JiIhqNhoSEBLy8vDAxMcHb25tjx46VxqUVQgghxHNAkmhRLMbGxty6dQuAhIQEjh07xrZt29i4cSN5eXn4+flhbm7Ozp07SUpKwszMDH9/f27dusUrr7yClZUVa9asUevLz89n9erVBAUFAZCWloa/vz+BgYEcOXKE1atXs2vXLsLCwrTiiI2NxcvLi0OHDjFkyBAGDx6sJrd79+4FYPv27WRkZLB27doiz+X9999nzZo1LF26lIMHD+Lk5ISfnx///vuvVrmxY8cSGxvL/v370dfXZ8CAAQ+9Rrm5uWRnZ2u9hBBCCPF8kiRaPJSiKGzfvp0tW7aovcampqYsWbKEBg0a0KBBA1avXk1BQQFLlizB3d0dV1dX4uLiOHv2LImJiejp6fHGG2/w1VdfqfUmJCSQmZlJYGAgAFOnTiUoKIiIiAicnZ3x9vZm7ty5LFu2jJs3b6rHdezYkSFDhuDk5ERUVBRVq1Zlx44dAFSrVg2AKlWqYGNjQ+XKlQudz7Vr11iwYAEff/wxAQEBuLm5sXjxYoyNjfn888+1ysbExODr64ubmxujR49m9+7dWrHcb+rUqVhaWqovOzs7Ha+6EEIIIco7SaJFkTZu3IiZmRlGRkYEBATQu3dvJkyYAIC7u7vWOOjDhw9z8uRJzM3NMTMzw8zMjMqVK3Pz5k3S0tIACAoKIjExkQsXLgCwYsUKOnXqhJWVlVpHfHy8eryZmRl+fn4UFBSQnp6utuXh4aG+12g02NjYcOnSpWKfV1paGnl5efj4+KjbKlasSLNmzUhNTdUqe29btra2AA9ta8yYMWRlZamvc+fOFTsuIYQQQjxb5MFCUaS2bduyYMECDAwMqFGjBvr6//dPxdTUVKtsTk4OTZo0YcWKFYXquds73LRpU+rWrcuqVasYPHgw69at05oxIycnh0GDBhEeHl6ojtq1a6vvK1asqLVPo9FQUFCg0zk+yr1taTQagIe2ZWhoiKGh4ROJRQghhBDliyTRokimpqY4OTkVq2zjxo1ZvXo11tbWWFhYPLBcUFAQK1asoFatWlSoUIFOnTpp1ZGSklLsNotyt3c8Pz//gWXq1q2LgYEBSUlJ2NvbA5CXl8e+ffuIiIjQuW0hhBBCvFhkOId4bEFBQVStWpWuXbuyc+dO0tPTSUxMJDw8nPPnz2uVO3jwIDExMbz++utavbZRUVHs3r2bsLAwkpOTOXHiBN99912hBwsfxtraGmNjYzZv3szff/9NVlZWoTKmpqYMHjyYUaNGsXnzZlJSUggNDeX69esMHDjw8S6EEEIIIV4YkkSLx2ZiYsIvv/xC7dq16dGjB66urgwcOJCbN29q9Uw7OTnRrFkzjhw5os7KcZeHhwc///wzx48fp1WrVnh6ehIdHU2NGjWKHYe+vj5z585l0aJF1KhRg65duxZZbtq0aQQGBvL222/TuHFjTp48yZYtW6hUqZJuF0AIIYQQLxyNoihKWQchxPMoOzsbS0tLGr63ED1D47IO54k48HHfsg5BCCGEKFV3v7+zsrIeOkxVeqKFEEIIIYQoIUmihRBCCCGEKCFJooUQQgghhCghSaKFEEIIIYQoIUmihRBCCCGEKCFJooUQQgghhCghSaKFEEIIIYQoIUmihRBCCCGEKCFJooUQQgghhCgh/bIOQIjn3S+T+zx0xSMhhBBCPHukJ1oIIYQQQogSkiRaCCGEEEKIEpIkWgghhBBCiBKSJFoIIYQQQogSkiRaCCGEEEKIEpIkWgghhBBCiBKSJFoIIYQQQogSknmihXjCzk1rgbmRXlmHIYQQQjw3akcfLesQpCdaCCGEEEKIkpIkWgghhBBCiBKSJFoIIYQQQogSkiRaCCGEEEKIEpIkWgghhBBCiBKSJFoIIYQQQogSkiRaCCGEEEKIEpIkWgghhBBCiBKSJFoIIYQQQogSkiRaCCGEEEKIEpIkWgghhBBCiBKSJFoIIYQQQogSkiRaPJBGo3noa8KECU+k3eDgYLp16/ZE6tZFfHw8VlZWZR2GEEIIIcoR/bIOQJRfGRkZ6vvVq1cTHR3NsWPH1G1mZmbqe0VRyM/PR19f/kkJIYQQ4vknPdHigWxsbNSXpaUlGo1G/fznn39ibm7Ojz/+SJMmTTA0NGTXrl0UFBQwdepUHBwcMDY2pmHDhnz77bdqnfn5+QwcOFDdX69ePT755BN1/4QJE1i6dCnfffed2uOdmJjI6dOn0Wg0fP3117Rq1QpjY2OaNm3K8ePH2bdvH15eXpiZmREQEMDly5e1zmPJkiW4urpiZGRE/fr1mT9/vrrvbr1r166lbdu2mJiY0LBhQ/bs2QNAYmIi/fv3Jysr64n3wAshhBDi2SHdhuKxjB49mpkzZ+Lo6EilSpWYOnUqy5cvZ+HChTg7O/PLL7/w1ltvUa1aNXx9fSkoKKBWrVp88803VKlShd27d/POO+9ga2tLr169iIyMJDU1lezsbOLi4gCoXLkyFy5cAGD8+PHMmTOH2rVrM2DAAN58803Mzc355JNPMDExoVevXkRHR7NgwQIAVqxYQXR0NPPmzcPT05NDhw4RGhqKqakp/fr1U89j7NixzJw5E2dnZ8aOHUufPn04efIk3t7ezJkzR6sX/t4e+Hvl5uaSm5urfs7Ozn4i11wIIYQQZU+SaPFYJk2aRIcOHYA7SeSUKVPYvn07LVu2BMDR0ZFdu3axaNEifH19qVixIhMnTlSPd3BwYM+ePXz99df06tULMzMzjI2Nyc3NxcbGplB7kZGR+Pn5ATBs2DD69OlDQkICPj4+AAwcOJD4+Hi1/Pjx44mNjaVHjx5qeykpKSxatEgriY6MjKRTp04ATJw4kQYNGnDy5Enq16+v1Qv/MFOnTtU6NyGEEEI8vySJFo/Fy8tLfX/y5EmuX7+uJtV33bp1C09PT/XzZ599xhdffMHZs2e5ceMGt27dolGjRsVqz8PDQ31fvXp1ANzd3bW2Xbp0CYBr166RlpbGwIEDCQ0NVcvcvn0bS0vLB9Zra2sLwKVLl6hfv36x4gIYM2YMI0aMUD9nZ2djZ2dX7OOFEEII8ezQOYnOzMzk22+/JS0tjVGjRlG5cmUOHjxI9erVqVmzZmnGKMoxU1NT9X1OTg4AmzZtKvRvwNDQEIBVq1YRGRlJbGwsLVu2xNzcnI8//pjffvutWO1VrFhRfa/RaIrcVlBQoBXP4sWLad68uVY9enp6j6z3bj3FZWhoqJ6nEEIIIZ5vOiXRR44coX379lhaWnL69GlCQ0OpXLkya9eu5ezZsyxbtqy04xTPADc3NwwNDTl79iy+vr5FlklKSsLb25shQ4ao29LS0rTKGBgYkJ+f/9jxVK9enRo1anDq1CmCgoJ0rqe04hFCCCHE80OnJHrEiBEEBwczY8YMzM3N1e0dO3bkzTffLLXgxLPF3NycyMhIhg8fTkFBAS+//DJZWVkkJSVhYWFBv379cHZ2ZtmyZWzZsgUHBwe+/PJL9u3bh4ODg1pPnTp12LJlC8eOHaNKlSqFhl6UxMSJEwkPD8fS0hJ/f39yc3PZv38/V69e1Rp68TB16tQhJyeHhIQEGjZsiImJCSYmJjrHJIQQQohnn05T3O3bt49BgwYV2l6zZk0uXrz42EGJZ9dHH33EuHHjmDp1Kq6urvj7+7Np0yY1SR40aBA9evSgd+/eNG/enCtXrmj1SgOEhoZSr149vLy8qFatGklJSTrHExISwpIlS4iLi8Pd3R1fX1/i4+O1kvZH8fb25t1336V3795Uq1aNGTNm6ByPEEIIIZ4PGkVRlJIeZG1tzZYtW/D09MTc3JzDhw/j6OjItm3bGDBgAOfOnXsSsQrxTMnOzsbS0pLfx7hibqT36AOEEEIIUSy1o48+sbrvfn9nZWVhYWHxwHI69UR36dKFSZMmkZeXB9x5EOvs2bNERUURGBioW8RCCCGEEEI8I3RKomNjY8nJycHa2pobN27g6+uLk5MT5ubmxMTElHaMQgghhBBClCs6PVhoaWnJtm3bSEpK4vDhw+Tk5NC4cWPat29f2vEJIYQQQghR7jzWYis+Pj7qSnGZmZmlEY8QQgghhBDlnk7DOaZPn87q1avVz7169aJKlSrUrFmTw4cPl1pwQgghhBBClEc6JdELFy5UlzPetm0b27Zt48cffyQgIIBRo0aVaoBCCCGEEEKUNzoN57h48aKaRG/cuJFevXrx6quvUqdOnULLKwshhBBCCPG80aknulKlSupc0Js3b1YfKFQURZZHFkIIIYQQzz2deqJ79OjBm2++ibOzM1euXCEgIACAQ4cO4eTkVKoBCiGEEEIIUd7olETPnj2bOnXqcO7cOWbMmIGZmRkAGRkZhZZwFuJFZzf614eueCSEEEKIZ49Oy34LIR6tuMuGCiGEEKL8KO73t87zRKelpTFnzhxSU1MBcHNzIyIiAkdHR12rFEIIIYQQ4pmg04OFW7Zswc3Njb179+Lh4YGHhwe//fYbbm5ubNu2rbRjFEIIIYQQolzRaTiHp6cnfn5+TJs2TWv76NGj2bp1KwcPHiy1AIV4VslwDiGEEOLZU9zvb516olNTUxk4cGCh7QMGDCAlJUWXKoUQQgghhHhm6JREV6tWjeTk5ELbk5OTsba2ftyYhBBCCCGEKNd0erAwNDSUd955h1OnTuHt7Q1AUlIS06dPZ8SIEaUaoBBCCCGEEOWNTmOiFUVhzpw5xMbGcuHCBQBq1KjBqFGjCA8PR6PRlHqgQjxr7o6paja9GfrGOk+EA0DSe0mlFJUQQgghHuaJTnGn0WgYPnw4w4cP57///gPA3Nxct0iFEEIIIYR4xjxW99ilS5c4duwYAPXr16datWqlEpQQQgghhBDlmU4PFv7333+8/fbb1KhRA19fX3x9falRowZvvfUWWVlZpR2jEEIIIYQQ5YpOSXRISAi//fYbmzZtIjMzk8zMTDZu3Mj+/fsZNGhQaccohBBCCCFEuaLTcI6NGzeyZcsWXn75ZXWbn58fixcvxt/fv9SCE0IIIYQQojzSqSe6SpUqWFpaFtpuaWlJpUqVHjsoIYQQQgghyjOdkugPP/yQESNGcPHiRXXbxYsXGTVqFOPGjSu14IQQQgghhCiPdBrOsWDBAk6ePEnt2rWpXbs2AGfPnsXQ0JDLly+zaNEitezBgwdLJ1IhhBBCCCHKCZ2S6G7dupVyGEIIIYQQQjw7SpxE5+fn07ZtWzw8PLCysnoCIQkhhBBCCFG+lXhMtJ6eHq+++ipXr159EvEIIYQQQghR7un0YOFLL73EqVOnSjsW8QzSaDSsX7++rMMoFW3atCEiIqKswxBCCCHEM0CnJHry5MlERkayceNGMjIyyM7O1nqJZ19wcDAajQaNRkPFihWpXr06HTp04IsvvqCgoEAtl5GRQUBAQBlGWnKJiYloNBoyMzO1tq9du5aPPvqobIISQgghxDNFpwcLO3bsCECXLl3QaDTqdkVR0Gg05Ofnl050okz5+/sTFxdHfn4+f//9N5s3b2bYsGF8++23bNiwAX19fWxsbMo6zFJTuXLlsg5BCCGEEM8InXqid+zYob5++ukn9XX3s3g+GBoaYmNjQ82aNWncuDEffPAB3333HT/++CPx8fFA4eEcUVFRuLi4YGJigqOjI+PGjSMvL0+r3smTJ2NtbY25uTkhISGMHj2aRo0aqfuDg4Pp1q0bM2fOxNbWlipVqjB06FCteq5evUrfvn2pVKkSJiYmBAQEcOLECXX/mTNn6Ny5M5UqVcLU1JQGDRrwww8/cPr0adq2bQtApUqV0Gg0BAcHA4WHc+Tm5hIVFYWdnR2GhoY4OTnx+eefl87FFUIIIcQzTaeeaF9f39KOQzwj2rVrR8OGDVm7di0hISGF9pubmxMfH0+NGjU4evQooaGhmJub8/777wOwYsUKYmJimD9/Pj4+PqxatYrY2FgcHBy06tmxYwe2trbs2LGDkydP0rt3bxo1akRoaChwJ9E+ceIEGzZswMLCgqioKDp27EhKSgoVK1Zk6NCh3Lp1i19++QVTU1NSUlIwMzPDzs6ONWvWEBgYyLFjx7CwsMDY2LjIc+3bty979uxh7ty5NGzYkPT0dP75558HXpvc3Fxyc3PVzzK0SQghhHh+6ZRE//LLLw/d37p1a52CEc+G+vXrc+TIkSL3ffjhh+r7OnXqEBkZyapVq9Qk+tNPP2XgwIH0798fgOjoaLZu3UpOTo5WPZUqVWLevHno6elRv359OnXqREJCAqGhoWrynJSUhLe3N3AnObezs2P9+vX07NmTs2fPEhgYiLu7OwCOjo5q3XeHbVhbWz9wmsbjx4/z9ddfs23bNtq3b1+ojqJMnTqViRMnPrSMEEIIIZ4POiXRbdq0KbTt3rHRMib6+XZ37HtRVq9ezdy5c0lLSyMnJ4fbt29jYWGh7j927BhDhgzROqZZs2aFhgE1aNAAPT099bOtrS1Hjx4FIDU1FX19fZo3b67ur1KlCvXq1SM1NRWA8PBwBg8ezNatW2nfvj2BgYF4eHgU+xyTk5PR09Mr0V2XMWPGMGLECPVzdnY2dnZ2xT5eCCGEEM8OncZEX716Vet16dIlNm/eTNOmTdm6dWtpxyjKmdTU1ELDLwD27NlDUFAQHTt2ZOPGjRw6dIixY8dy69atErdRsWJFrc8ajUZrVpBHCQkJ4dSpU7z99tscPXoULy8vPv3002If/6AhHg9jaGiIhYWF1ksIIYQQzyedkmhLS0utV9WqVenQoQPTp09Xb9uL59NPP/3E0aNHCQwMLLRv9+7d2NvbM3bsWLy8vHB2dubMmTNaZerVq8e+ffu0tt3/+VFcXV25ffs2v/32m7rtypUrHDt2DDc3N3WbnZ0d7777LmvXrmXkyJEsXrwYAAMDA+Dhd0zc3d0pKCjg559/LlFsQgghhHgx6JREP0j16tU5duxYaVYpylBubi4XL17kr7/+4uDBg0yZMoWuXbvy2muv0bdv30LlnZ2dOXv2LKtWrSItLY25c+eybt06rTLvvfcen3/+OUuXLuXEiRNMnjyZI0eOPHB4SFGcnZ3p2rUroaGh7Nq1i8OHD/PWW29Rs2ZNunbtCkBERARbtmwhPT2dgwcPsmPHDlxdXQGwt7dHo9GwceNGLl++XGg8NtwZz92vXz8GDBjA+vXrSU9PJzExka+//rokl1AIIYQQzymdkugjR45ovQ4fPszmzZt59913taYqE8+2zZs3Y2trS506dfD392fHjh3MnTuX7777Tmu88l1dunRh+PDhhIWF0ahRI3bv3s24ceO0ygQFBTFmzBgiIyNp3Lgx6enpBAcHY2RkVKLY4uLiaNKkCa+99hotW7ZEURR++OEHdRhIfn4+Q4cOxdXVFX9/f1xcXJg/fz4ANWvWZOLEiYwePZrq1asTFhZWZBsLFizg9ddfZ8iQIdSvX5/Q0FCuXbtWojiFEEII8XzSKIqilPSgChUqoNFouP/QFi1a8MUXX1C/fv1SC1A8/zp06ICNjQ1ffvllWYdSqrKzs7G0tKTZ9GboG+v0DK8q6b2kUopKCCGEEA9z9/s7Kyvroc836fTNnp6ervW5QoUKVKtWrcS9ieLFc/36dRYuXIifnx96enqsXLmS7du3s23btrIOTQghhBCi2HRKou3t7Us7DvGC0Gg0/PDDD8TExHDz5k3q1avHmjVr1LmYhRBCCCGeBTrfY05ISGD27NnqvLyurq5ERERIMiQeytjYmO3bt5d1GEIIIYQQj0WnBwvnz5+Pv78/5ubmDBs2jGHDhmFhYUHHjh357LPPSjtGIYQQQgghyhWdeqKnTJnC7NmztWY1CA8Px8fHhylTpjB06NBSC1AIIYQQQojyRqee6MzMTPz9/Qttf/XVV8nKynrsoIQQQgghhCjPdEqiu3TpUmgRDYDvvvuO11577bGDEkIIIYQQojzTaTiHm5sbMTExJCYm0rJlSwB+/fVXkpKSGDlyJHPnzlXLhoeHl06kQgghhBBClBM6Lbbi4OBQvMo1Gk6dOlXioIR4HshiK0IIIcSzp7iLreiURAshHq24/xMKIYQQovwo7ve3TmOii8vCwkJ6ooUQQgghxHPniSbR0skthBBCCCGeR080iRZCCCGEEOJ5JEm0EEIIIYQQJSRJtBBCCCGEECX0RJNojUbzJKsXQgghhBCiTMiDhUIIIYQQQpSQTitA7Nixg7Zt2z6y3I8//kjNmjV1aUKI58Yu/wBM9R9vsZX7+f7yc6nWJ4QQQoiS0akn2t/fn7p16zJ58mTOnTv3wHIvv/wyhoaGOgcnhBBCCCFEeaRTEv3XX38RFhbGt99+i6OjI35+fnz99dfcunWrtOMTQgghhBCi3NEpia5atSrDhw8nOTmZ3377DRcXF4YMGUKNGjUIDw/n8OHDpR2nEEIIIYQQ5cZjP1jYuHFjxowZQ1hYGDk5OXzxxRc0adKEVq1a8ccff5RGjEIIIYQQQpQrOifReXl5fPvtt3Ts2BF7e3u2bNnCvHnz+Pvvvzl58iT29vb07NmzNGMVQgghhBCiXNBpyoD33nuPlStXoigKb7/9NjNmzOCll15S95uamjJz5kxq1KhRaoEKIYQQQghRXuiURKekpPDpp5/So0ePB86+UbVqVXbs2PFYwQkhhBBCCFEelXg4R15eHvb29rRo0eKh09fp6+vj6+v7WMEJIYQQQghRHpU4ia5YsSJr1qx5ErEIIYQQQgjxTNDpwcJu3bqxfv36Ug5FCCGEEEKIZ4NOY6KdnZ2ZNGkSSUlJNGnSBFNTU6394eHhpRKcEEIIIYQQ5ZFGURSlpAc5ODg8uEKNhlOnTj1WUKJ8iY+PJyIigszMzDJp//Tp0zg4OHDo0CEaNWpUJjHoIjs7G0tLSza19MZUX6e/Vx/I95efS7U+IYQQQtxx9/s7KysLCwuLB5bTaThHenr6A1+SQD9dwcHBaDQa9VWlShX8/f05cuRIqbXRu3dvjh8/rvPx+fn5TJs2jfr162NsbEzlypVp3rw5S5YsKdbxdnZ2ZGRkaE2jKIQQQghRlh57xUJR9vz9/cnIyCAjI4OEhAT09fV57bXXSq1+Y2NjrK2tdT5+4sSJzJ49m48++oiUlBR27NjBO++8U+yebT09PWxsbNAv5d5cIYQQQghd6ZxEnz9/nvnz5zN69GhGjBih9RJPl6GhITY2NtjY2NCoUSNGjx7NuXPnuHz5slomKioKFxcXTExMcHR0ZNy4ceTl5an7Dx8+TNu2bTE3N8fCwoImTZqwf/9+4M5wDisrK602v//+e5o2bYqRkRFVq1ale/fuD4xvw4YNDBkyhJ49e+Lg4EDDhg0ZOHAgkZGRapmCggJmzJiBk5MThoaG1K5dm5iYGODOcA6NRkNycrJa/vfffycgIAAzMzOqV6/O22+/zT///KPub9OmDeHh4bz//vtUrlwZGxsbJkyYoBVXZmYmgwYNonr16hgZGfHSSy+xceNGdf+uXbto1aoVxsbG2NnZER4ezrVr1x79AxFCCCHEc0+nrr2EhAS6dOmCo6Mjf/75Jy+99BKnT59GURQaN25c2jGKEsjJyWH58uU4OTlRpUoVdbu5uTnx8fHUqFGDo0ePEhoairm5Oe+//z4AQUFBeHp6smDBAvT09EhOTqZixYpFtrFp0ya6d+/O2LFjWbZsGbdu3eKHH354YEw2Njb89NNPDBkyhGrVqhVZZsyYMSxevJjZs2fz8ssvk5GRwZ9//llk2czMTNq1a0dISAizZ8/mxo0bREVF0atXL3766Se13NKlSxkxYgS//fYbe/bsITg4GB8fHzp06EBBQQEBAQH8999/LF++nLp165KSkoKenh4AaWlp+Pv7M3nyZL744gsuX75MWFgYYWFhxMXFFRlXbm4uubm56ufs7OwHXhMhhBBCPNt0erCwWbNmBAQEMHHiRMzNzTl8+DDW1tYEBQXh7+/P4MGDn0SsogjBwcEsX74cIyMjAK5du4atrS0bN2586B80M2fOZNWqVWpvs4WFBZ9++in9+vUrVPb+Bwu9vb1xdHRk+fLlxYoxJSWF119/nWPHjtGgQQO8vb3p2rUrAQEBAPz3339Uq1aNefPmERISUuj4+x8snDx5Mjt37mTLli1qmfPnz2NnZ8exY8dwcXGhTZs25Ofns3PnTrVMs2bNaNeuHdOmTWPr1q0EBASQmpqKi4tLoTZDQkLQ09Nj0aJF6rZdu3bh6+vLtWvX1Ot9rwkTJjBx4sRC2+XBQiGEEOLZ8UQfLExNTaVv377AnZUJb9y4gZmZGZMmTWL69Om6RSx01rZtW5KTk0lOTmbv3r34+fkREBDAmTNn1DKrV6/Gx8cHGxsbzMzM+PDDDzl79qy6f8SIEYSEhNC+fXumTZtGWlraA9tLTk7mlVdeKXZ8bm5u/P777/z6668MGDCAS5cu0blzZzVhTk1NJTc3t9h1Hj58mB07dmBmZqa+6tevD6AVt4eHh9Zxtra2XLp0ST2HWrVqFZlA320jPj5eqw0/Pz8KCgpIT08v8pgxY8aQlZWlvs6dO1es8xFCCCHEs0enJNrU1JRbt24BdxKTexOXe8eliqfD1NQUJycnnJycaNq0KUuWLOHatWssXrwYgD179hAUFETHjh3ZuHEjhw4dYuzYserPEO70ov7xxx906tSJn376CTc3N9atW1dke8bGxiWOsUKFCjRt2pSIiAjWrl1LfHw8n3/+Oenp6SWuLycnh86dO6t/ONx9nThxgtatW6vl7h+OotFoKCgoKNY55OTkMGjQIK36Dx8+zIkTJ6hbt26RxxgaGmJhYaH1EkIIIcTzSad7zC1atGDXrl24urrSsWNHRo4cydGjR1m7di0tWrQo7RhFCWk0GipUqMCNGzcA2L17N/b29owdO1Ytc28v9V0uLi64uLgwfPhw+vTpQ1xcXJEPDHp4eJCQkED//v11jtHNzQ24M/zE2dkZY2NjEhISihzOcb/GjRuzZs0a6tSpo/OMHR4eHpw/f57jx48X2RvduHFjUlJScHJy0ql+IYQQQjzfdMpAZs2aRU5ODnBn+rKcnBxWr16Ns7Mzs2bNKtUAxaPl5uZy8eJFAK5evcq8efPU3lq4s8Lk2bNnWbVqFU2bNmXTpk1avcw3btxg1KhRvP766zg4OHD+/Hn27dtHYGBgke2NHz+eV155hbp16/LGG29w+/ZtfvjhB6Kiooos//rrr+Pj44O3tzc2Njakp6czZswYXFxcqF+/Pvr6+kRFRfH+++9jYGCAj48Ply9f5o8//mDgwIGF6hs6dCiLFy+mT58+6uwbJ0+eZNWqVSxZskR9OPBhfH19ad26NYGBgcyaNQsnJyf+/PNPNBoN/v7+REVF0aJFC8LCwggJCcHU1JSUlBS2bdvGvHnzHlm/EKL8y8/P15qlSAjxYqhYsWKxcoVH0SmJdnR0VN+bmpqycOHCxw5E6G7z5s3Y2toCd2bhqF+/Pt988w1t2rQBoEuXLgwfPpywsDByc3Pp1KkT48aNU6d809PT48qVK/Tt25e///6bqlWr0qNHjyIfkoM708d98803fPTRR0ybNg0LCwutYRT38/PzY+XKlUydOpWsrCxsbGxo164dEyZMUHuSx40bh76+PtHR0Vy4cAFbW1vefffdIuurUaMGSUlJREVF8eqrr5Kbm4u9vT3+/v5UqFD8EUpr1qwhMjKSPn36cO3aNZycnJg2bRpwp6f6559/ZuzYsbRq1QpFUahbty69e/cudv1CiPJJURQuXrxYZquwCiHKnpWVFTY2Nmg0Gp3r0Gl2jnvl5OSo40zvkrGgQsiy30KUVxkZGWRmZmJtbY2JicljfYkKIZ4tiqJw/fp1Ll26hJWVldoJea/izs6h0zd7eno6YWFhJCYmcvPmTa3ANBoN+fn5ulQrhBBCPFH5+flqAn3vXPpCiBfH3ckFLl26hLW1tc5DO3RKot966y0UReGLL76gevXq8le8EEKIZ8LdMdAmJiZlHIkQoizd/R2Ql5f3dJPow4cPc+DAAerVq6dTo0IIIURZks4fIV5spfE7QKd5ops2bSoLSQghhBBCiBeWTj3RS5Ys4d133+Wvv/7ipZdeKrSoxf0rxQkhhBBCCPE80akn+vLly6SlpdG/f3+aNm1Ko0aN8PT0VP8rhBBCiAeLj4/HysrqidWfmJiIRqMptWn82rRpQ0RERKnUVRy6xj9hwgT1Wa3169c/kdgepk6dOsyZM+eptyvKhk490QMGDMDT05OVK1fKg4VCCCFeCMHBwWRmZpZJclbW1q5dW+iu85Pk7e1NRkYGlpaWxT4mNTWViRMnsm7dOlq0aEGlSpWKfeyNGzeoWrUqhw8ffqyVavft24epqanOx4tni05J9JkzZ9iwYYMsiSyEEEK8ACpXrvxU2zMwMMDGxqZEx6SlpQHQtWvXEnfubdu2DXt7e53zmlu3bmFgYEC1atV0Ov5x3W2/vFEUhfz8fHVhteeNTsM52rVrx+HDh0s7FiGEEKLMffvtt7i7u2NsbEyVKlVo3749o0aNYunSpXz33XdoNBo0Gg2JiYlFDjtITk5Go9Fw+vRpdVt8fDy1a9fGxMSE7t27c+XKFXXf6dOnqVChAvv379eKY86cOdjb2xda0KwoP/zwAy4uLhgbG9O2bVuttu/atWsXrVq1wtjYGDs7O8LDw7l27Zq6f/78+Tg7O2NkZET16tV5/fXX1X33D+fIyMigU6dOGBsb4+DgwFdffVVoKINGo2HJkiV0794dExMTnJ2d2bBhwyPPBQoP57g7/GXLli24urpiZmaGv78/GRkZwJ1hHJ07dwagQoUKahJdUFDApEmTqFWrFoaGhjRq1IjNmzcXau+7776jS5cual2NGjVi0aJF2NnZYWJiQq9evcjKylLLBwcH061bN2JiYqhRo4Y6W1lR12DRokW89tprmJiY4Orqyp49ezh58iRt2rTB1NQUb29v9Q8AuPPHQNeuXalevTpmZmY0bdqU7du3a8Vbp04dPvroI/r27YuFhQXvvPMO7dq1IywsTKvc5cuXMTAwICEh4ZHX/G6dffr0wdTUlJo1a/LZZ5+p+0+fPo1GoyE5OVndlpmZqf6/cO/P7ccff6RJkyYYGhqya9euYl3TR/2sbt26RVhYGLa2thgZGWFvb8/UqVO1YgkJCaFatWpYWFg8nVxV0cGiRYsUOzs7Zfz48cq3336rfPfdd1ovIYSiZGVlKYCSlZVV1qEIIf6/GzduKCkpKcqNGzeK3H/hwgVFX19fmTVrlpKenq4cOXJE+eyzz5T//vtP6dWrl+Lv769kZGQoGRkZSm5urrJjxw4FUK5evarWcejQIQVQ0tPTFUVRlF9//VWpUKGCMn36dOXYsWPKJ598olhZWSmWlpbqMR06dFCGDBmiFYuHh4cSHR39yHM6e/asYmhoqIwYMUL5888/leXLlyvVq1fXiuvkyZOKqampMnv2bOX48eNKUlKS4unpqQQHByuKoij79u1T9PT0lK+++ko5ffq0cvDgQeWTTz5R2/D19VWGDRumfm7fvr3SqFEj5ddff1UOHDig+Pr6KsbGxsrs2bPVMoBSq1Yt5auvvlJOnDihhIeHK2ZmZsqVK1ceeU73X9e4uDilYsWKSvv27ZV9+/YpBw4cUFxdXZU333xTURRF+e+//5S4uDgFUH8+iqIos2bNUiwsLJSVK1cqf/75p/L+++8rFStWVI4fP662lZ+fr1hbWyu7d+9WFEVRxo8fr5iamirt2rVTDh06pPz888+Kk5OT2paiKEq/fv0UMzMz5e2331Z+//135ffff1cURVHs7e0LXYOaNWsqq1evVo4dO6Z069ZNqVOnjtKuXTtl8+bNSkpKitKiRQvF399fPSY5OVlZuHChcvToUeX48ePKhx9+qBgZGSlnzpxRy9jb2ysWFhbKzJkzlZMnTyonT55UVqxYoVSqVEm5efOmWm7WrFlKnTp1lIKCgkdec3t7e8Xc3FyZOnWqcuzYMWXu3LmKnp6esnXrVkVRFCU9PV0BlEOHDqnHXL16VQGUHTt2aP3cPDw8lK1btyonT55Urly5Uqxr+qif1ccff6zY2dkpv/zyi3L69Gll586dyldffaUe3759e6Vz587Kvn37lOPHjysjR45UqlSp8sB/bw/7XVDc72+dkmiNRvPAV4UKFXSpUojnjiTRQpQ/j0qiDxw4oADK6dOnC+3r16+f0rVrV61txUmi+/Tpo3Ts2FHruN69e2sl0atXr9ZKgA4cOKBoNBq1jocZM2aM4ubmprUtKipKK66BAwcq77zzjlaZnTt3KhUqVFBu3LihrFmzRrGwsFCys7OLbOPeJDo1NVUBlH379qn7T5w4oQCFEsgPP/xQ/ZyTk6MAyo8//vjIcyoqiQaUkydPqmU+++wzpXr16urndevWKff3DdaoUUOJiYnR2ta0aVOtP1iSkpIUa2trJT8/X1GUO0m0np6ecv78ebXMjz/+qFSoUEFNzvv166dUr15dyc3N1aq7qCT63muwZ88eBVA+//xzddvKlSsVIyOjh16PBg0aKJ9++qlWO926ddMqc+PGDaVSpUrK6tWr1W0eHh7KhAkTHlr3vXXem8wryp1/pwEBAYqilCyJXr9+vVY9xbmmj/pZvffee0q7du2K/INg586dioWFhdYfEIqiKHXr1lUWLVpU5PmWRhKt03COgoKCB75kyW8hhBDPqoYNG/LKK6/g7u5Oz549Wbx4MVevXn2sOlNTU2nevLnWtpYtW2p97tatG3p6eqxbtw64M3yhbdu21KlTp1TqP3z4MPHx8ZiZmakvPz8/CgoKSE9Pp0OHDtjb2+Po6Mjbb7/NihUruH79epHtHTt2DH19fRo3bqxuc3JyKvJBvnunvDU1NcXCwoJLly498pyKYmJiQt26ddXPtra2D60rOzubCxcu4OPjo7Xdx8eH1NRU9fN3333Ha6+9RoUK/5cS1a5dm5o1a6qfW7ZsSUFBAceOHVO3ubu7F2sc8r3XoHr16uqx9267efMm2dnZAOTk5BAZGYmrqytWVlaYmZmRmprK2bNnter18vLS+mxkZMTbb7/NF198AcDBgwf5/fffCQ4OfmSM957n/Z/vvVbFdX9s8PBrWpyfVXBwMMnJydSrV4/w8HC2bt2qljt8+DA5OTlUqVJF6994enq61lCZ0qZTEi2EEEI8j/T09Ni2bRs//vgjbm5ufPrpp9SrV4/09PQiy99NvBRFUbfdXVq8JAwMDOjbty9xcXHcunWLr776igEDBuh2EkXIyclh0KBBJCcnq6/Dhw9z4sQJ6tati7m5OQcPHmTlypXY2toSHR1Nw4YNH3uKvPtn9NBoNMUa413cuu697rrasGGDOh66JIo7C8e9cd8dq13UtrvXJTIyknXr1jFlyhR27txJcnIy7u7u3Lp165Hth4SEsG3bNs6fP09cXBzt2rXD3t6+ZCf2ACX5t/4kZihp3Lgx6enpfPTRR9y4cYNevXqp4/ZzcnKwtbXV+vednJzMsWPHGDVqVKnHclexH5ecO3cu77zzDkZGRsydO/ehZcPDwx87MCGEEKIsaDQafHx88PHxITo6Gnt7e9atW4eBgUGhu613Z2PIyMhQe2LvffAKwNXVld9++01r26+//lqo3ZCQEF566SXmz5/P7du36dGjR7HidXV1LfTA3v31N27cmJSUlIfOPqGvr0/79u1p374948ePx8rKip9++qlQHPXq1eP27dscOnSIJk2aAHDy5MnH7rEvbRYWFtSoUYOkpCR8fX3V7UlJSTRr1gyAEydOcObMGTp06KB17NmzZ7lw4QI1atQA7lzPChUqqA8QPklJSUkEBwfTvXt34E6CWNSDokVxd3fHy8uLxYsX89VXXzFv3rwStX3/v5tff/0VV1dXQPvf+t01Qe7/t/4wD7umxflZwZ2fae/evenduzevv/46/v7+/PvvvzRu3JiLFy+ir69frLs3paXYSfTs2bMJCgrCyMiI2bNnP7CcRqORJFoIIcQz6bfffiMhIYFXX30Va2trfvvtNy5fvoyrqys3b95ky5YtHDt2jCpVqmBpaYmTkxN2dnZMmDCBmJgYjh8/TmxsrFad4eHh+Pj4MHPmTLp27cqWLVuKnCHC1dWVFi1aEBUVxYABAzA2Ni5WzO+++y6xsbGMGjWKkJAQDhw4QHx8vFaZqKgoWrRoQVhYGCEhIZiampKSksK2bduYN28eGzdu5NSpU7Ru3ZpKlSrxww8/UFBQUGTSWL9+fdq3b88777zDggULqFixIiNHjsTY2LjcrRsxatQoxo8fT926dWnUqBFxcXEkJyezYsUK4M5Qjvbt22NiYqJ1nJGREf369WPmzJlkZ2cTHh5Or169Sjztni6cnZ1Zu3YtnTt3RqPRMG7cuBL13oeEhBAWFoapqamaiBdXUlISM2bMoFu3bmzbto1vvvmGTZs2AWBsbEyLFi2YNm0aDg4OXLp0iQ8//LDYdT/qmj7qZzVr1ixsbW3x9PSkQoUKfPPNN9jY2GBlZUX79u1p2bIl3bp1Y8aMGbi4uHDhwgU2bdpE9+7dixxeUhqKPZwjPT2dKlWqqO8f9Dp16tQTCVQIIYR40iwsLPjll1/o2LEjLi4ufPjhh8TGxhIQEEBoaCj16tXDy8uLatWqkZSURMWKFVm5ciV//vknHh4eTJ8+ncmTJ2vV2aJFCxYvXswnn3xCw4YN2bp16wOTj4EDB3Lr1q0SDeWoXbs2a9asYf369TRs2JCFCxcyZcoUrTIeHh78/PPPHD9+nFatWuHp6Ul0dLTaK2hlZcXatWtp164drq6uLFy4kJUrV9KgQYMi21y2bBnVq1endevWdO/endDQUMzNzTEyMip23E9DeHg4I0aMYOTIkbi7u7N582Y2bNiAs7MzoD213b2cnJzo0aMHHTt25NVXX8XDw4P58+c/lZhnzZpFpUqV8Pb2pnPnzvj5+WmNP3+UPn36oK+vT58+fUr88xg5ciT79+/H09OTyZMnM2vWLPz8/NT9X3zxBbdv36ZJkyZEREQU+rf+MI+6po/6WZmbmzNjxgy8vLxo2rQpp0+f5ocfflCnNPzhhx9o3bo1/fv3x8XFhTfeeIMzZ86o49CfBI1SwgFFeXl51K9fn40bN6pd/EKIwrKzs7G0tCQrKwsLC4uyDkcIAdy8eZP09HQcHBzKXcIH8NFHH/HNN99w5MiRsg6lRM6fP4+dnR3bt2/nlVdeKetwiuWff/7B1taW8+fPayVaEyZMYP369SUaqlCenD59mrp167Jv374SJd916tQhIiLiiSzvXh6v6cN+FxT3+7vES8hUrFiRmzdvljxaIV5Qiz74EWNDE8JiO5d1KEKIcuruuNd58+aVqHevrPz000/k5OTg7u5ORkYG77//PnXq1KF169ZlHVqx/fvvv8yaNeuJ9lQ+TXl5eVy5coUPP/yQFi1alCiBFrrRaXaOoUOHMn36dG7fvl3a8QghhBAvnLCwMJo0aUKbNm0KDeV49913tabtuvf17rvvlkm8eXl5fPDBBzRo0IDu3btTrVo1EhMTC82g8SDl4ZxcXFx47733nkpbT0NSUhK2trbs27ePhQsXau3buXPnA6+3mZlZGUX87CvxcA6A7t27k5CQgJmZGe7u7oWmMlm7dm2pBSjEs+ru7aAZQ1dJT7QQ5UR5H85RlEuXLqlzCN/PwsICa2vrpxzR43sez6k8u3HjBn/99dcD9z9s1pbnVZkM54A7DyAEBgbqcqgQQgghSsDa2vq5Syqfx3Mqz4yNjV/IRPlJ0ymJjouLK+04hBBCCCGEeGbolETfdenSJXUJzHr16slflUIIIYQQ4oWg04OF2dnZvP3229SsWRNfX198fX2pWbMmb731FllZWaUdoxBCCCGEEOWKTkl0aGgov/32Gxs3biQzM5PMzEw2btzI/v37GTRoUGnHKIQQQgghRLmi03COjRs3smXLFl5++WV1m5+fH4sXL8bf37/UghNCCCGEEKI80imJrlKlCpaWloW2W1paUqlSpccOSgghhHjWNBm17Km2d+Djvk+1vfImODiYzMxM1q9fX9ahiBeUTsM5PvzwQ0aMGMHFixfVbRcvXmTUqFGMGzeu1IITQgghROkIDg5Go9Gg0WgwMDDAycmJSZMmlfuF006fPo1Goym0ZPQnn3xCfHx8mcQkBOjYE71gwQJOnjxJ7dq1qV27NgBnz57F0NCQy5cvs2jRIrXswYMHSydSIYQQQjwWf39/4uLiyM3N5YcffmDo0KFUrFiRMWPGaJW7desWBgYGZRRl8RR1R1yIp0mnnuhu3boRGRnJ2LFjefvtt3n77bcZO3YskZGRdO3aVeslRFl7UC+GEEK8aAwNDbGxscHe3p7BgwfTvn17NmzYQHBwMN26dSMmJoYaNWpQr149AM6dO0evXr2wsrKicuXKdO3aldOnTwOwdetWjIyMyMzM1Gpj2LBhtGvXTv28a9cuWrVqhbGxMXZ2doSHh3Pt2jV1f506dZgyZQoDBgzA3Nyc2rVr87///U/d7+DgAICnpycajYY2bdoAqDHflZubS3h4ONbW1hgZGfHyyy+zb98+dX9iYiIajYaEhAS8vLwwMTHB29tbnapXiJIqUU/0qVOncHR0ZPz48U8qHlGOXb58mejoaDZt2sTff/9NpUqVaNiwIdHR0fj4+JR1eA9kZ2dHRkYGVatWLetQhBCiXDE2NubKlSsAJCQkYGFhwbZt2wDIy8vDz8+Pli1bsnPnTvT19Zk8eTL+/v4cOXKEV155BSsrK9asWcPAgQMByM/PZ/Xq1cTExACQlpaGv78/kydP5osvvuDy5cuEhYURFhamtXBbbGwsH330ER988AHffvstgwcPxtfXl3r16rF3716aNWvG9u3badCgwQN7yN9//33WrFnD0qVLsbe3Z8aMGfj5+XHy5EkqV66slhs7diyxsbFUq1aNd999lwEDBpCUlPRErq94vpWoJ9rDw4OXXnqJDz74gL179z6pmEQ5FRgYyKFDh1i6dCnHjx9nw4YNtGnTRv0FXBL5+fkUFBQ8gSgL09PTw8bGBn39x1pbSAghnhuKorB9+3a2bNmi9hqbmpqyZMkSGjRoQIMGDVi9ejUFBQUsWbIEd3d3XF1diYuL4+zZsyQmJqKnp8cbb7zBV199pdabkJBAZmYmgYGBAEydOpWgoCAiIiJwdnbG29ubuXPnsmzZMm7evKke17FjR4YMGYKTkxNRUVFUrVqVHTt2AFCtWjXgzqQGNjY2WgnxXdeuXWPBggV8/PHHBAQE4ObmxuLFizE2Nubzzz/XKhsTE4Ovry9ubm6MHj2a3bt3a8UiRHGVKIn+559/mDp1KpcuXaJLly7Y2toSGhrK999/L/8An3OZmZns3LmT6dOn07ZtW+zt7WnWrBljxoyhS5cuaplBgwZRvXp1jIyMeOmll9i4cSMA8fHxWFlZsWHDBtzc3DA0NOTs2bPk5uYSGRlJzZo1MTU1pXnz5iQmJmq1/bi3Au8fzlHcW3qTJ0/G2toac3NzQkJCGD16NI0aNSr9iyuEEE/Jxo0bMTMzw8jIiICAAHr37s2ECRMAcHd31+rlPXz4MCdPnsTc3BwzMzPMzMyoXLkyN2/eJC0tDYCgoCASExO5cOECACtWrKBTp05YWVmpdcTHx6vHm5mZ4efnR0FBAenp6WpbHh4e6nuNRoONjQ2XLl0q9nmlpaWRl5endVe0YsWKNGvWjNTUVK2y97Zla2sLUKK2hLirREm0kZERnTt3ZsmSJWRkZLBmzRqqVKmi/tXYrVs39XaNeL7c/eW3fv16cnNzC+0vKCggICCApKQkli9fTkpKCtOmTUNPT08tc/36daZPn86SJUv4448/sLa2JiwsjD179rBq1SqOHDlCz5498ff358SJE8D/3QoMDAzkyJEjrF69ml27dhEWFqbVfmxsLF5eXhw6dIghQ4YwePDgR45zu3tLb//+/ejr6zNgwAB134oVK4iJiWH69OkcOHCA2rVrs2DBgofWl5ubS3Z2ttZLCCHKk7Zt25KcnMyJEye4ceMGS5cuxdTUFED97105OTk0adKE5ORkrdfx48d58803AWjatCl169Zl1apV3Lhxg3Xr1hEUFKRVx6BBg7SOP3z4MCdOnKBu3bpquYoVK2q1rdFontjdynvb0mg0AE/tzqh4vuh8f1uj0eDt7Y23tzfTpk3jxIkTbNiwgfj4eAYPHsysWbMYOnRoacYqypC+vj7x8fGEhoaycOFCGjdujK+vL2+88QYeHh5s376dvXv3kpqaiouLCwCOjo5adeTl5TF//nwaNmwI3JnR5e6twRo1agAQGRnJ5s2biYuLY8qUKVq3AgGcnZ2ZO3cuvr6+LFiwACMjI+D/bgUCREVFMXv2bHbs2KE+HFOUu7f0AEaPHk2nTp24efMmRkZGfPrppwwcOJD+/fsDEB0dzdatW8nJyXlgfVOnTmXixIklvbRCCPHUmJqa4uTkVKyyjRs3ZvXq1VhbW2NhYfHAckFBQaxYsYJatWpRoUIFOnXqpFVHSkpKsdssyt3e8fz8/AeWqVu3LgYGBiQlJWFvbw/c+c7Zt2+f+v0hRGnTaXaOojg7OzNy5Eh++eUXLly4wKuvvlpaVYtyIjAwkAsXLrBhwwb8/f1JTEykcePGxMfHk5ycTK1atdQEuigGBgZat9GOHj1Kfn4+Li4uWrf6fv75Z/VW4ZO8FfiwW3rHjh2jWbNmWuXv/3y/MWPGkJWVpb7OnTv30PJCCFGeBQUFUbVqVbp27crOnTtJT08nMTGR8PBwzp8/r1Xu4MGDxMTE8Prrr2NoaKjui4qKYvfu3YSFhak94N99912hu4kPY21tjbGxMZs3b+bvv/8mKyurUBlTU1MGDx7MqFGj2Lx5MykpKYSGhnL9+nX1oUchSpvOPdFpaWnMmTNHHWvk5uZGREQEjo6OVKlShSpVqpRakKL8MDIyokOHDnTo0IFx48YREhLC+PHjiYyMfOSxxsbG6q0zuHObT09PjwMHDmgN+4A7w0fulhk0aBDh4eGF6rs7RznodiuwtG/pGRoaan15CCFeLM/bCoImJib88ssvREVF0aNHD/777z9q1qzJK6+8otUz7eTkRLNmzdi7dy9z5szRqsPDw4Off/6ZsWPH0qpVKxRFoW7duvTu3bvYcejr6zN37lwmTZpEdHQ0rVq1KvTsDMC0adMoKCjg7bff5r///sPLy4stW7bISsriidEpid6yZQtdunShUaNG6iD+pKQk3Nzc+P777+nQoUOpBinKLzc3N9avX4+Hhwfnz5/n+PHjD+2Nvpenpyf5+flcunSJVq1aFVmmNG4F6qJevXrs27ePvn3/70vx3vlGhRDiWfOw1f0etM/GxoalS5c+su7ffvvtgfuaNm3K1q1bH7j/7rzT97p/Xv+QkBBCQkK0tt0fs5GREXPnzmXu3LlFttOmTRsURdHa1qhRo0LbhCgunZLo0aNHM3z4cKZNm1Zoe1RUlCTRz6ErV67Qs2dPBgwYgIeHB+bm5uzfv58ZM2bQtWtXfH19ad26NYGBgcyaNQsnJyf+/PNPNBoN/v7+Rdbp4uJCUFAQffv2JTY2Fk9PTy5fvkxCQgIeHh506tSJqKgoWrRoQVhYGCEhIZiampKSksK2bduYN2/eEzvf9957j9DQULy8vPD29mb16tUcOXKk0DhvIYQQQryYdEqiU1NT+frrrwttHzBgQKFbOeL5YGZmRvPmzZk9e7Y6lZCdnR2hoaF88MEHAKxZs4bIyEj69OnDtWvXcHJyKvSH1v3i4uKYPHkyI0eO5K+//qJq1aq0aNGC1157DSidW4G6CAoK4tSpU0RGRnLz5k169epFcHCwzI8uhBBCCAA0ig73Mezs7Jg1axY9e/bU2v71118TGRnJ2bNnSy1AIcqLDh06YGNjw5dfflms8tnZ2VhaWjJj6CqMDU0Ii+38hCMUQjzKzZs3SU9Px8HBQZ3dRwjx4nnY74K7399ZWVkPnZlGp57o0NBQ3nnnHU6dOoW3tzdwZ0z09OnTGTFihC5VClGuXL9+nYULF+Ln54eenh4rV65k+/bt6nK4QgghhHix6ZREjxs3DnNzc2JjYxkzZgwANWrUYMKECUXOoiDEs0aj0fDDDz8QExPDzZs3qVevHmvWrKF9+/ZlHZoQQgghyoFiJ9EbNmwgICCAihUrotFoGD58OMOHD+e///4DwNzc/IkFKcTTZmxszPbt28s6DCGEEEKUU8VebKV79+5kZmYCoKenpy5KYW5uLgm0EEIIIYR4oRQ7ia5WrRq//vorAIqiaC2aIYQQQgghxIuk2MM53n33Xbp27YpGo1GXVX6Qh61vL4QQQgghxLOu2En0hAkTeOONNzh58iRdunQhLi4OKyurJxiaEEII8ew4O8n9qbZXO/poscs+6u7x+PHjmTBhwmNGVFhwcDCZmZmsX7++1OvWRXx8PBEREerwVCEeR4lm56hfvz7169dn/Pjx9OzZExMTkycVlxBCCCFKSUZGhvp+9erVREdHc+zYMXWbmZmZ+l5RFPLz89HX12kCLyFeGMUeE32v8ePHSwIthBBCPCNsbGzUl6WlpTos08bGhj///BNzc3N+/PFHmjRpgqGhIbt27aKgoICpU6fi4OCAsbExDRs25Ntvv1XrzM/PZ+DAger+evXq8cknn6j7J0yYwNKlS/nuu+/UoaCJiYmcPn0ajUbD119/TatWrTA2NqZp06YcP36cffv24eXlhZmZGQEBAVy+fFnrPJYsWYKrqytGRkbUr1+f+fPnq/vu1rt27Vratm2LiYkJDRs2ZM+ePQAkJibSv39/srKy1HieRO+7eHHo9Gfm33//TWRkJAkJCVy6dIn7Fz2UMdFC/J9BUwIeuuKREEKUB6NHj2bmzJk4OjpSqVIlpk6dyvLly1m4cCHOzs788ssvvPXWW1SrVg1fX18KCgqoVasW33zzDVWqVGH37t2888472Nra0qtXLyIjI0lNTSU7O5u4uDgAKleuzIULF4A7HXJz5syhdu3aDBgwgDfffBNzc3M++eQTTExM6NWrF9HR0SxYsACAFStWEB0dzbx58/D09OTQoUOEhoZiampKv3791PMYO3YsM2fOxNnZmbFjx9KnTx9OnjyJt7c3c+bM0eqFv7cHXoiS0imJDg4O5uzZs4wbNw5bW1uZqUMIIYR4xk2aNIkOHToAkJuby5QpU9i+fTstW7YEwNHRkV27drFo0SJ8fX2pWLEiEydOVI93cHBgz549fP311/Tq1QszMzOMjY3Jzc0tcjKCyMhI/Pz8ABg2bBh9+vQhISEBHx8fAAYOHEh8fLxafvz48cTGxtKjRw+1vZSUFBYtWqSVREdGRtKpUycAJk6cSIMGDTh58iT169fX6oUX4nHplETv2rWLnTt30qhRo1IORwghhBBlwcvLS31/8uRJrl+/ribVd926dQtPT0/182effcYXX3zB2bNnuXHjBrdu3Sp2buDh4aG+r169OgDu7u5a2+6uSXHt2jXS0tIYOHAgoaGhapnbt29jaWn5wHptbW0BuHTpEvXr1y9WXEIUl05JtJ2dXaEhHEIIIYR4dpmamqrvc3JyANi0aRM1a9bUKmdoaAjAqlWriIyMJDY2lpYtW2Jubs7HH3/Mb7/9Vqz2KlasqL6/e0f7/m0FBQVa8SxevJjmzZtr1aOnp/fIeu/WI0Rp0imJnjNnDqNHj2bRokXUqVOnlEMSQgghRFlyc3PD0NCQs2fP4uvrW2SZpKQkvL29GTJkiLotLS1Nq4yBgUGpPCdVvXp1atSowalTpwgKCtK5ntKKRwjQMYnu3bs3169fp27dupiYmGj91Qfw77//lkpwQgghhHj6zM3NiYyMZPjw4RQUFPDyyy+TlZVFUlISFhYW9OvXD2dnZ5YtW8aWLVtwcHDgyy+/ZN++fTg4OKj11KlThy1btnDs2DGqVKlSaOhFSUycOJHw8HAsLS3x9/cnNzeX/fv3c/XqVUaMGFGsOurUqUNOTg4JCQk0bNgQExMTmW1M6EznnmghhBBCPL8++ugjqlWrxtSpUzl16hRWVlY0btyYDz74AIBBgwZx6NAhevfujUajoU+fPgwZMoQff/xRrSM0NJTExES8vLzIyclhx44dOt/BDgkJwcTEhI8//phRo0ZhamqKu7s7ERERxa7D29ubd999l969e3PlypUntsiMeDFoFBncLMQTkZ2djaWlJVlZWTLFnRDlxM2bN0lPT8fBwQEjI6OyDkcIUUYe9ruguN/fxe6Jzs7OVivKzs5+aFlJGIQQQgghxPOs2El0pUqVyMjIwNraGisrqyLnhlYUBY1GI4P2hbjHx6FvY3TfcwO6GLv820cXEkIIIcRTUewk+qeffqJy5coA7Nix44kFJIQQQgghRHlX7CT63iluHjTdzf2GDBnCpEmTqFq1askjE0IIIYQQopyq8CQrX758+SPHTwshhBBCCPGseaJJtEz8IYQQQgghnkdPNIkWQgghhBDieSRJtBBCCCGEECUkSbQQQgghhBAlJEm0EEIIIVQajYb169eXdRilok2bNiVaFlyIkij2FHe6eOutt2T1QiGEEC8En099nmp7Se8llah8cHAwS5cuBUBfX5/KlSvj4eFBnz59CA4OpkKFO/1qGRkZVKpUqdTjfZISExNp27YtV69excrKSt2+du1aKpbCYldCFEXnJDozM5O9e/dy6dIlCgoKtPb17dsXgAULFjxedEIIIYQoNf7+/sTFxZGfn8/ff//N5s2bGTZsGN9++y0bNmxAX18fGxubsg6z1NxdJE6IJ0Gn4Rzff/89tWvXxt/fn7CwMIYNG6a+5LaJALmFJoQQ5ZGhoSE2NjbUrFmTxo0b88EHH/Ddd9/x448/Eh8fDxQezhEVFYWLiwsmJiY4Ojoybtw48vLytOqdPHky1tbWmJubExISwujRo2nUqJG6Pzg4mG7dujFz5kxsbW2pUqUKQ4cO1arn6tWr9O3bl0qVKmFiYkJAQAAnTpxQ9585c4bOnTtTqVIlTE1NadCgAT/88AOnT5+mbdu2AFSqVAmNRkNwcDBQ+LsoNzeXqKgo7OzsMDQ0xMnJic8//7x0Lq544eiURI8cOZIBAwaQk5NDZmYmV69eVV///vtvacf43Lp8+TKDBw+mdu3a6i82Pz8/kpJKdouuLCUmJqLRaMjMzNTavnbtWj766KMn3v7Fixd57733cHR0xNDQEDs7Ozp37kxCQsITb1sIIZ4H7dq1o2HDhqxdu7bI/ebm5sTHx5OSksInn3zC4sWLmT17trp/xYoVxMTEMH36dA4cOEDt2rWLvBO9Y8cO0tLS2LFjB0uXLiU+Pl5N3OFOor1//342bNjAnj17UBSFjh07qon20KFDyc3N5ZdffuHo0aNMnz4dMzMz7OzsWLNmDQDHjh0jIyODTz75pMhz6du3LytXrmTu3LmkpqayaNEizMzMdL104gWn03COv/76i/DwcExMTEo7nhdKYGAgt27dYunSpTg6OvL333+TkJDAlStXyjq0Yrm/J+JeT+MW2unTp/Hx8cHKyoqPP/4Yd3d38vLy2LJlC0OHDuXPP//Uqd78/Hw0Go06PlAIIZ539evX58iRI0Xu+/DDD9X3derUITIyklWrVvH+++8D8OmnnzJw4ED69+8PQHR0NFu3biUnJ0ernkqVKjFv3jz09PSoX78+nTp1IiEhgdDQUE6cOMGGDRtISkrC29sbuJOc29nZsX79enr27MnZs2cJDAzE3d0dAEdHR7Xuu9851tbWWmOi73X8+HG+/vprtm3bRvv27QvVIURJ6ZQl+Pn5sX///tKO5YWSmZnJzp07mT59Om3btsXe3p5mzZoxZswYunTpAtxJEjUaDcnJyVrHaTQaEhMTgf/rCd60aRMeHh4YGRnRokULfv/9d/WY+Ph4rKysWL9+Pc7OzhgZGeHn58e5c+e0YlqwYAF169bFwMCAevXq8eWXX2rt12g0LFiwgC5dumBqakpoaGixbqF98MEHNG/evNA1aNiwIZMmTVI/L1myBFdXV4yMjKhfvz7z589/6DUcMmQIGo2GvXv3EhgYiIuLCw0aNGDEiBH8+uuvarlZs2bh7u6OqakpdnZ2DBkyROuX+93rs2HDBtzc3DA0NOTs2bPk5uYSGRlJzZo1MTU1pXnz5up1F0KI54miKGg0miL3rV69Gh8fH2xsbDAzM+PDDz/k7Nmz6v5jx47RrFkzrWPu/wzQoEED9PT01M+2trZcunQJgNTUVPT19bW+K6pUqUK9evVITU0FIDw8nMmTJ+Pj48P48eMfmPQ/SHJyMnp6evj6+pboOCEeRKckulOnTowaNYoJEyawZs0aNmzYoPUSj2ZmZoaZmRnr168nNzf3sesbNWoUsbGx7Nu3j2rVqtG5c2etnuLr168TExPDsmXLSEpKIjMzkzfeeEPdv27dOoYNG8bIkSP5/fffGTRoEP3792fHjh1a7UyYMIHu3btz9OhRJk6cWKxbaEFBQezdu5e0tDR12x9//MGRI0d48803gTs9DtHR0cTExJCamsqUKVMYN26c+iT5/f799182b97M0KFDMTU1LbT/3p6IChUqMHfuXP744w+WLl3KTz/9pPag3Ht9pk+fzpIlS/jjjz+wtrYmLCyMPXv2sGrVKo4cOULPnj3x9/fXGqN3r9zcXLKzs7VeQgjxLEhNTcXBwaHQ9j179hAUFETHjh3ZuHEjhw4dYuzYsdy6davEbdw/S4ZGoyk0McHDhISEcOrUKd5++22OHj2Kl5cXn376abGPNzY2LnZZIYpDpyQ6NDSUc+fOMWnSJHr27Em3bt3UV/fu3Us7xueSvr4+8fHxLF26FCsrK3x8fPjggw9K/Jf1XePHj6dDhw64u7uzdOlS/v77b9atW6fuz8vLY968ebRs2ZImTZqwdOlSdu/ezd69ewGYOXMmwcHBDBkyBBcXF0aMGEGPHj2YOXOmVjtvvvkm/fv3x9HREXt7e61baDY2NlhaWhaKrUGDBjRs2JCvvvpK3bZixQqaN2+Ok5OTGn9sbCw9evTAwcGBHj16MHz4cBYtWlTk+Z48eRJFUahfv/4jr01ERARt27alTp06tGvXjsmTJ/P1119rlcnLy2P+/Pl4e3tTr149/vnnH+Li4vjmm29o1aoVdevWJTIykpdffpm4uLgi25k6dSqWlpbqy87O7pGxCSFEWfvpp584evQogYGBhfbt3r0be3t7xo4di5eXF87Ozpw5c0arTL169di3b5/Wtvs/P4qrqyu3b9/mt99+U7dduXKFY8eO4ebmpm6zs7Pj3XffZe3atYwcOZLFixcDYGBgANwZjvcg7u7uFBQU8PPPP5coNiEeRKckuqCg4IGvh/0DFtoCAwO5cOECGzZswN/fn8TERBo3bqz1oEVxtWzZUn1fuXJlrVtgcCdpb9q0qfq5fv36WFlZqWVSU1Px8dGe49THx0erDgAvL68SxwZ3eqPvJtGKorBy5UqCgoIAuHbtGmlpaQwcOFDtoTczM2Py5Mlavdf3UhSl2G1v376dV155hZo1a2Jubs7bb7/NlStXuH79ulrGwMAADw8P9fPRo0fJz8/HxcVFK6aff/75gTGNGTOGrKws9XX/cBkhhChrubm5XLx4kb/++ouDBw8yZcoUunbtymuvvaZOT3svZ2dnzp49y6pVq0hLS2Pu3LlaHTQA7733Hp9//jlLly7lxIkTTJ48mSNHjjxweEhRnJ2d6dq1K6GhoezatYvDhw/z1ltvUbNmTbp27Qrc6RDZsmUL6enpHDx4kB07duDq6gqAvb09Go2GjRs3cvny5ULjseHOeO5+/foxYMAA1q9fT3p6OomJiYU6VYQorie62Ip4NCMjIzp06ECHDh0YN24cISEhjB8/Xmvi+3sTxoc9zPc0FDV0ojj69OlDVFQUBw8e5MaNG5w7d47evXsDqL/sFi9eXGjs9L3j5+7l7OyMRqN55MODp0+f5rXXXmPw4MHExMRQuXJldu3axcCBA7l165b6cKyxsbHWL/ycnBz09PQ4cOBAoRge9CS3oaEhhoaGD41HCCHK0ubNm7G1tUVfX59KlSrRsGFD5s6dS79+/Yp8mLpLly4MHz6csLAwcnNz6dSpE+PGjWPChAlqmaCgIE6dOkVkZCQ3b96kV69eBAcHq3c6iysuLo5hw4bx2muvcevWLVq3bs0PP/ygDgPJz89n6NChnD9/HgsLC/z9/dVZQmrWrMnEiRMZPXo0/fv3p2/fvkV2SC1YsIAPPviAIUOGcOXKFWrXrs0HH3xQojiFuEunJHru3LlFbtdoNBgZGeHk5ETr1q0fmACJB3Nzc1Pn56xWrRpwZ/UoT09PAK2HDO/166+/Urt2beDOXJvHjx9X/0IHuH37Nvv371cf9jh27BiZmZlqGVdXV5KSkujXr596TFJSktZttKIU5xYaQK1atfD19WXFihXcuHGDDh06YG1tDUD16tWpUaMGp06dUnunH6Vy5cr4+fnx2WefER4eXii5z8zMxMrKigMHDlBQUEBsbKz6BVGcXgdPT0/y8/O5dOkSrVq1KlZMQogXW0lXEHza7p9S7kHuv9M3Y8YMZsyYobXt/nUAxo0bx7hx49TPHTp0UIfr3W37fnPmzNH6XKlSJZYtW/bAuB41/vn+GIBCD4MbGRkxa9YsZs2a9dC6hCgOnZLo2bNnc/nyZa5fv64uDXr16lVMTEwwMzPj0qVLODo6smPHDhkX+gBXrlyhZ8+eDBgwAA8PD8zNzdm/fz8zZsxQb10ZGxvTokULpk2bhoODA5cuXdKaauhekyZNokqVKlSvXp2xY8dStWpVunXrpu6vWLEi7733HnPnzkVfX5+wsDBatGihJtWjRo2iV69eeHp60r59e77//nvWrl3L9u3bH3oe995C69ixI8bGxg/sqQ0KCmL8+PHcunVLa45RgIkTJxIeHo6lpSX+/v7k5uayf/9+rl69yogRI4qs77PPPsPHx4dmzZoxadIkPDw8uH37Ntu2bWPBggWkpqbi5OREXl4en376KZ07dyYpKYmFCxc+9JwAXFxcCAoKom/fvsTGxuLp6cnly5dJSEjAw8ODTp06PbIOIYR4EVy/fp2FCxfi5+eHnp4eK1euZPv27Wzbtq2sQxPiidJpTPSUKVNo2rQpJ06c4MqVK1y5coXjx4/TvHlzPvnkE86ePYuNjQ3Dhw8v7XifG2ZmZjRv3pzZs2fTunVrXnrpJcaNG0doaCjz5s1Ty33xxRfcvn2bJk2aEBERweTJk4usb9q0aQwbNowmTZpw8eJFvv/+e7WXGMDExISoqCjefPNNfHx8MDMzY/Xq1er+bt268cknnzBz5kwaNGjAokWLiIuLo02bNg89j3tvoVWvXp2wsLAHln399dfVscj3Jvhw56nrJUuWEBcXh7u7O76+vsTHxxf5tPhdjo6OHDx4kLZt2zJy5EheeuklOnToQEJCgjrRf8OGDZk1axbTp0/npZdeYsWKFUydOvWh53RXXFwcffv2ZeTIkdSrV49u3bqxb98+tcdfCCHEnbvQP/zwA61bt6ZJkyZ8//33rFmzRp2LWYjnlUYpyRNa/1/dunVZs2aN1pKeAIcOHSIwMJBTp06xe/duAgMDycjIKK1YRRESExNp27YtV69efeAE8/Hx8URERBRaVVA8WdnZ2VhaWvJhry4Y3Te1ky7GLv+2FKIS4sV28+ZN0tPTcXBwwMjIqKzDEUKUkYf9Lrj7/Z2VlYWFhcUD69CpJzojI4Pbt28X2n779m0uXrwIQI0aNfjvv/90qV4IIYQQQohyTackum3btgwaNIhDhw6p2w4dOsTgwYNp164dcGeKsIfdihdCCCGEEOJZpVMS/fnnn1O5cmWaNGmiTuvl5eVF5cqV+fzzz4E7Y35jY2NLNVhRWJs2bVAU5YFDOQCCg4NlKIcQQgghRCnSaXYOGxsbtm3bxp9//snx48eBOysW1atXTy3Ttm3b0olQCCGEEEKIcuaxFlupX79+sZZdFkIIIYQQ4nlS7CR6xIgRfPTRR5iamj5w3t67ZBJzIYQQQgjxPCt2En3o0CF1yel7Hyi8371LJwshhBDixVHWU6qePn0aBwcHDh06VGgaXiFKW7GT6B07dhT5XgghhBDwc2vfp9qe7y8/l6h8cHAwS5cuVT9XrlyZpk2bMmPGDDw8PEolpt69e9OxY0edj8/Pz+fjjz8mPj6eM2fOYGxsjLOzM6GhoYSEhDzyeDs7OzIyMqhatarOMQhRXDrNznG/7Oxs1q9fz59//lka1QkhhBDiCfD39ycjI4OMjAwSEhLQ19fntddeK7X6jY2Nsba21vn4iRMnMnv2bD766CNSUlLYsWMH77zzTrF7tvX09LCxsUFf/7Ee+RKiWHT6V9arVy9at25NWFgYN27cwMvLi9OnT6MoCqtWrSIwMLC04xTimTVq8ZcPXfFICCGeFkNDQ2xsbIA7M22NHj2aVq1acfnyZapVqwZAVFQU69at4/z589jY2BAUFER0dDQV///Kq4cPHyYiIoL9+/ej0WhwdnZm0aJFeHl5FTmc4/vvv2fSpEkcPXoUMzMzWrVqxbp164qMb8OGDQwZMoSePXuq2xo2bKhVpqCggJkzZ/K///2Pc+fOUb16dQYNGsTYsWOLHM7x+++/M2rUKHbu3ImpqSmvvvoqs2fPVnur27Rpg4eHB0ZGRixZsgQDAwPeffddJkyYoLaZmZlJVFQU69evJysrCycnJ6ZNm6b+AbJr1y7GjBnD/v37qVq1Kt27d2fq1KmYmprq/sMS5Z5OPdG//PILrVq1AmDdunUoikJmZiZz585l8uTJpRqgEEIIIUpfTk4Oy5cvx8nJiSpVqqjbzc3NiY+PJyUlhU8++YTFixcze/ZsdX9QUBC1atVi3759HDhwgNGjR6sJ9v02bdpE9+7d6dixI4cOHSIhIYFmzZo9MCYbGxt++uknLl++/MAyY8aMYdq0aYwbN46UlBS++uorqlevXmTZzMxM2rVrh6enJ/v372fz5s38/fff9OrVS6vc0qVLMTU15bfffmPGjBlMmjSJbdu2AXeS9oCAAJKSkli+fDkpKSlMmzYNPT09ANLS0vD39ycwMJAjR46wevVqdu3aRVhY2APPQTwfNIqiKCU9yNjYmOPHj2NnZ0ffvn2pUaMG06ZN4+zZs7i5uZGTk/MkYhXimZKdnY2lpSVZWVnSEy1EOXHz5k3S09NxcHDAyMioVOt+FsZEL1++XD3va9euYWtry8aNG2ncuPEDj5s5cyarVq1i//79AFhYWPDpp5/Sr1+/QmXv74n29vbG0dGR5cuXFyvGlJQUXn/9dY4dO0aDBg3w9vama9euBAQEAPDff/9RrVo15s2bV+QY6ft7oidPnszOnTvZsmWLWub8+fPY2dlx7NgxXFxcaNOmDfn5+ezcuVMt06xZM9q1a8e0adPYunUrAQEBpKam4uLiUqjNkJAQ9PT0WLRokbpt165d+Pr6cu3atVL/dyZKx8N+FxT3+1unnmg7Ozv27NnDtWvX2Lx5M6+++ioAV69elX8sQgghRDnVtm1bkpOTSU5OZu/evfj5+REQEMCZM2fUMqtXr8bHxwcbGxvMzMz48MMPOXv2rLp/xIgRhISE0L59e6ZNm0ZaWtoD20tOTuaVV14pdnxubm78/vvv/PrrrwwYMIBLly7RuXNnNWFOTU0lNze32HUePnyYHTt2YGZmpr7urm9xb9z3P1hpa2vLpUuX1HOoVatWkQn03Tbi4+O12vDz86OgoID09PRin7t49uiUREdERKi3c2xtbWnTpg1wZ5iHu7t7acYnhBBCiFJiamqKk5MTTk5ONG3alCVLlnDt2jUWL14MwJ49ewgKCqJjx45s3LiRQ4cOMXbsWG7duqXWMWHCBP744w86derETz/9hJub2wPHOBsbG5c4xgoVKtC0aVMiIiJYu3Yt8fHxfP7556Snp5e4vpycHDp37qz+4XD3deLECVq3bq2Wu384ikajoaCgoFjnkJOTw6BBg7TqP3z4MCdOnKBu3bolilc8W3R6sHDIkCE0a9aMc+fO0aFDBypUuJOLOzo6yphoIYQQ4hmh0WioUKECN27cAGD37t3Y29szduxYtcy9vdR3ubi44OLiwvDhw+nTpw9xcXF07969UDkPDw8SEhLo37+/zjG6ubkBd4afODs7Y2xsTEJCQrGmvGvcuDFr1qyhTp06Os/Y4eHhwfnz5zl+/HiRvdGNGzcmJSUFJycnneoXzy6dp7jz8vKiU6dO/PXXX9y+fRuATp064ePjU2rBCSGEEKL05ObmcvHiRS5evEhqairvvfee2lsL4OzszNmzZ1m1ahVpaWnMnTtXq5f5xo0bhIWFkZiYyJkzZ0hKSmLfvn24uroW2d748eNZuXIl48ePJzU1laNHjzJ9+vQHxvf6668ze/ZsfvvtN86cOUNiYiJDhw7FxcWF+vXrY2RkRFRUFO+//z7Lli0jLS2NX3/9lc8//7zI+oYOHcq///5Lnz592LdvH2lpaWzZsoX+/fuTn59frGvm6+tL69atCQwMZNu2baSnp/Pjjz+yefNm4M5sJrt37yYsLEzt5f7uu+/kwcIXgE5J9PXr1xk4cCAmJiY0aNBAHSv13nvvMW3atFINUAghhBClY/Pmzdja2mJra0vz5s3Zt28f33zzjToss0uXLgwfPpywsDAaNWrE7t27GTdunHq8np4eV65coW/fvri4uNCrVy8CAgKYOHFike21adOGb775hg0bNtCoUSPatWvH3r17Hxifn58f33//PZ07d8bFxYV+/fpRv359tm7dqvYkjxs3jpEjRxIdHY2rqyu9e/dWxy/fr0aNGiQlJZGfn8+rr76Ku7s7ERERWFlZqXfRi2PNmjU0bdqUPn364Obmxvvvv68m4R4eHvz8888cP36cVq1a4enpSXR0NDVq1Ch2/eLZpNPsHMOGDSMpKYk5c+bg7+/PkSNHcHR05LvvvmPChAkPXRZciBeFzM4hRPnzJGfnEEI8O0pjdg6dBgitX7+e1atX06JFCzQajbq9QYMGD31KV4gX0bGPf8bM6MlMuO86tt0TqVcIIYQQD6fTcI7Lly8XuazntWvXtJJqIYQQQgghnkc6JdFeXl5s2rRJ/Xw3cV6yZAktW7YsnciEEEIIIYQop3QazjFlyhQCAgJISUnh9u3bfPLJJ6SkpLB7925+/rlkKygJIYQQQgjxrNGpJ/rll18mOTmZ27dv4+7uztatW7G2tmbPnj00adKktGMUQgghhBCiXNFt5nGgbt266gpHQgghhBBCvEiKnURnZ2cXu1KZzksIIYQQQjzPip1EW1lZFXvmjeKuAiSEEEIIIcSzqNhJ9I4dO9T3p0+fZvTo0QQHB6uzcezZs4elS5cyderU0o9SCCGEEEKIcqTYSbSvr6/6ftKkScyaNYs+ffqo27p06YK7uzv/+9//6NevX+lGKYQQQogX1unTp3FwcODQoUM0atSorMMRAtDxwcI9e/awcOHCQtu9vLwICQl57KCEEEKIZ828kd8/1fbCYjuX+JjLly8THR3Npk2b+Pvvv6lUqRINGzYkOjoaHx+fJxBl6bCzsyMjI4OqVauWdShCqHSa4s7Ozq7ImTmWLFmCnZ3dYwclRGmIj4/HysqqrMMQQohyIzAwkEOHDrF06VKOHz/Ohg0baNOmDVeuXClxXfn5+RQUFDyBKAvT09PDxsYGfX2dJxUTotTplETPnj2bTz/9FHd3d0JCQggJCcHDw4NPP/2U2bNnl3aM4gV37tw5BgwYQI0aNTAwMMDe3p5hw4Zp/dKvU6cOc+bMKbsghRCinMvMzGTnzp1Mnz6dtm3bYm9vT7NmzRgzZgxdunRRywwaNIjq1atjZGTESy+9xMaNG4H/65jYsGEDbm5uGBoacvbsWXJzc4mMjKRmzZqYmprSvHlzEhMTtdretWsXrVq1wtjYGDs7O8LDw7l27Zq6v06dOkyZMoUBAwZgbm5O7dq1+d///qfuP336NBqNhuTkZAASExPRaDQkJCTg5eWFiYkJ3t7eHDt2TKvdyZMnY21tjbm5OSEhIYwePVqGg4hSo1MS3bFjR06cOEGXLl34999/+ffff+ncuTPHjx+nY8eOpR2jeIGdOnUKLy8vTpw4wcqVKzl58iQLFy4kISGBli1b8u+//z71mPLy8p56m0II8bjMzMwwMzNj/fr15ObmFtpfUFBAQEAASUlJLF++nJSUFKZNm4aenp5a5vr160yfPp0lS5bwxx9/YG1tTVhYGHv27GHVqlUcOXKEnj174u/vz4kTJwBIS0vD39+fwMBAjhw5wurVq9m1axdhYWFa7cfGxuLl5cWhQ4cYMmQIgwcPLpQU32/s2LHExsayf/9+9PX1GTBggLpvxYoVxMTEMH36dA4cOEDt2rVZsGDB41xCIbTolEQD1KpVi5iYGNauXcvatWuJiYkpNJRjyJAh/PPPP48dpHhxDR06FAMDA7Zu3Yqvry+1a9cmICCA7du389dffzF27FjatGnDmTNnGD58OBqNptBUjFu2bMHV1RUzMzP8/f3JyMjQ2r9kyRJcXV0xMjKifv36zJ8/X913t/dj9erV+Pr6YmRkxIoVK57KuQshRGnS19cnPj6epUuXYmVlhY+PDx988AFHjhwBYPv27ezdu5e1a9fSoUMHHB0dee211wgICFDryMvLY/78+Xh7e1OvXj3++ecf4uLi+Oabb2jVqhV169YlMjKSl19+mbi4OACmTp1KUFAQERERODs74+3tzdy5c1m2bBk3b95U6+7YsSNDhgzBycmJqKgoqlatqjUzWFFiYmLw9fXFzc2N0aNHs3v3brXOTz/9lIEDB9K/f39cXFyIjo7G3d29tC+reIHpnEQXx/Lly0u0SIsQ9/r333/ZsmULQ4YMwdjYWGufjY0NQUFBrF69mjVr1lCrVi0mTZpERkaGVpJ8/fp1Zs6cyZdffskvv/zC2bNniYyMVPevWLGC6OhoYmJiSE1NZcqUKYwbN46lS5dqtTd69GiGDRtGamoqfn5+Rcabm5tLdna21ksIIcqTwMBALly4wIYNG/D39ycxMZHGjRsTHx9PcnIytWrVwsXF5YHHGxgY4OHhoX4+evQo+fn5uLi4qD3dZmZm/Pzzz6SlpQFw+PBh4uPjtfb7+flRUFBAenq6Wte99Wo0GmxsbLh06dJDz+feY2xtbQHUY44dO0azZs20yt//WYjH8URH6CuK8iSrF8+5EydOoCgKrq6uRe53dXXl6tWr5Ofno6enh7m5OTY2Nlpl8vLyWLhwIXXr1gUgLCyMSZMmqfvHjx9PbGwsPXr0AMDBwYGUlBQWLVqkNVVjRESEWuZBpk6dysSJE3U6VyGEeFqMjIzo0KEDHTp0YNy4cYSEhDB+/HitDoYHMTY21rrbl5OTg56eHgcOHNAa9gF3ho/cLTNo0CDCw8ML1Ve7dm31fcWKFbX2aTSaRz64eO8xd+N6Wg87CiGPuYpy73H+GDMxMVETaLjTU3G3l+LatWukpaUxcOBAQkND1TK3b9/G0tJSqx4vL69HtjVmzBhGjBihfs7OzpbZaoQQ5Z6bmxvr16/Hw8OD8+fPc/z48Yf2Rt/L09OT/Px8Ll26RKtWrYos07hxY1JSUnBycirNsB+pXr167Nu3j759+6rb9u3b91RjEM83SaJFueXk5IRGoyE1NZXu3bsX2p+amkqlSpWoVq3aA+soqmfjblKek5MDwOLFi2nevLlWuft7VExNTR8Zr6GhIYaGho8sJ4QQZeHKlSv07NmTAQMG4OHhgbm5Ofv372fGjBl07doVX19fWrduTWBgILNmzcLJyYk///wTjUaDv79/kXW6uLgQFBRE3759iY2NxdPTk8uXL5OQkICHhwedOnUiKiqKFi1aEBYWRkhICKampqSkpLBt2zbmzZv3xM73vffeIzQ0FC8vL7y9vVm9ejVHjhzB0dHxibUpXiySRItyq0qVKnTo0IH58+czfPhwrXHRFy9eZMWKFfTt2xeNRoOBgQH5+fklqr969erUqFGDU6dOERQUVNrhCyFeMLosfvI0mZmZ0bx5c2bPnk1aWhp5eXnY2dkRGhrKBx98AMCaNWuIjIykT58+XLt2DScnJ6ZNm/bQeuPi4pg8eTIjR47kr7/+omrVqrRo0YLXXnsNuDNu+eeff2bs2LG0atUKRVGoW7cuvXv3fqLnGxQUxKlTp4iMjOTmzZv06tWL4OBg9u7d+0TbFS8OjfIEBy6bm5tz+PBh+atP6OzEiRN4e3vj6urK5MmTcXBw4I8//mDUqFHk5uby66+/UrlyZV599VWMjY2ZP38+hoaGVK1alfj4eCIiIsjMzFTrW79+Pd27d1d7o5csWUJ4eDjTpk3D39+f3Nxc9u/fz9WrVxkxYsRjLTWbnZ2NpaUlez/cgJnRo3uydeE6tt0TqVeI59XNmzdJT0/HwcEBIyOjsg5HPGUdOnTAxsaGL7/8sqxDEWXsYb8L7n5/Z2VlYWFh8cA6nujsHG+99dZDGxfiUZydndm/fz+Ojo706tWLunXr8s4779C2bVv27NlD5cqVAZg0aRKnT5+mbt26Dx3ecb+QkBCWLFlCXFwc7u7u+Pr6Eh8fj4ODw5M6JSGEEE/B9evXmTVrFn/88Qd//vkn48ePZ/v27VoPjQvxOHTqia5Tpw4DBgwgODhY68laIcT/kZ5oIcof6Yl+cdy4cYPOnTtz6NAhbt68Sb169fjwww8fOdOSeDGUWU90REQEa9euxdHRkQ4dOrBq1aoiVz8SQgghhCgLxsbGbN++nStXrnDt2jUOHjwoCbQoVTon0cnJyezduxdXV1fee+89bG1tCQsL4+DBg6UdoxBCCCGEEOXKY42Jbty4MXPnzuXChQuMHz+eJUuW0LRpUxo1asQXX3whi60IIYQQQojn0mNNcZeXl8e6deuIi4tj27ZttGjRgoEDB3L+/Hk++OADtm/fzldffVVasQohhBBCCFEu6JREHzx4kLi4OFauXEmFChXo27cvs2fPpn79+mqZ7t2707Rp01ILVAghhBBCiPJCpyS6adOmdOjQgQULFtCtW7dCq8IBODg48MYbbzx2gEIIIYQQQpQ3OiXRp06dwt7e/qFlTE1NiYuL0ykoIYQQQgghyjOdHix8VAIthBBCCPG0tGnThoiIiLIOQ7xgit0TXalSJTQaTbHK/vvvvzoHJIQQQjyLYt56/am2N3b5tyUqf/nyZaKjo9m0aRN///03lSpVomHDhkRHR+Pj4/OEoixdiYmJtG3blqtXr2JlZaVuX7t2bZFDS0vbxYsXiYmJYdOmTfz1119YW1vTqFEjIiIieOWVV554+6J8KXYSPWfOHPX9lStXmDx5Mn5+frRs2RKAPXv2sGXLFsaNG1fqQQrxLKs3yvehKx4JIcTTEBgYyK1bt1i6dCmOjo78/fffJCQkcOXKlbIOrVjy8vIeuK9y5cpPvP3Tp0/j4+ODlZUVH3/8Me7u7uTl5bFlyxaGDh3Kn3/+qVO9+fn5aDQaKlR4rFmHRRko9k+sX79+6ispKYlJkyaxcuVKwsPDCQ8PZ+XKlUyaNImff/75ScYrhBBCiBLKzMxk586dTJ8+nbZt22Jvb0+zZs0YM2YMXbp0Ae4kiRqNhuTkZK3jNBoNiYmJwJ2eYI1Gw6ZNm/Dw8MDIyIgWLVrw+++/q8fEx8djZWXF+vXrcXZ2xsjICD8/P86dO6cV04IFC6hbty4GBgbUq1ePL7/8Umu/RqNhwYIFdOnSBVNTU0JDQ2nbti3wf3fHg4ODAe3hHB988AHNmzcvdA0aNmzIpEmT1M9LlizB1dUVIyMj6tevz/z58x96DYcMGYJGo2Hv3r0EBgbi4uJCgwYNGDFiBL/++qtabtasWbi7u2NqaoqdnR1DhgwhJyen0PXZsGEDbm5uGBoacvbsWXJzc4mMjKRmzZqYmprSvHlz9bqL8kmnP3u2bNmCv79/oe3+/v5s3779sYMSQgghROkxMzPDzMyM9evXk5ub+9j1jRo1itjYWPbt20e1atXo3LmzVk/x9evXiYmJYdmyZSQlJZGZmak1Y9e6desYNmwYI0eO5Pfff2fQoEH079+fHTt2aLUzYcIEunfvztGjR5k4cSJr1qwB4NixY2RkZPDJJ58Uii0oKIi9e/eSlpambvvjjz84cuQIb775JgArVqwgOjqamJgYUlNTmTJlCuPGjWPp0qVFnu+///7L5s2bGTp0KKampoX23zu0pEKFCsydO5c//viDpUuX8tNPP/H+++9rlb9+/TrTp09nyZIl/PHHH1hbWxMWFsaePXtYtWoVR44coWfPnvj7+3PixIkH/RhEGdMpia5SpQrfffddoe3fffcdVapUeeyghBBCCFF69PX1iY+PZ+nSpVhZWeHj48MHH3zAkSNHdKpv/PjxdOjQAXd3d5YuXcrff//NunXr1P15eXnMmzePli1b0qRJE5YuXcru3bvZu3cvADNnziQ4OJghQ4bg4uLCiBEj6NGjBzNnztRq580336R///44Ojpib2+vDtuwtrbGxsYGS0vLQrE1aNCAhg0bai32tmLFCpo3b46Tk5Maf2xsLD169MDBwYEePXowfPhwFi1aVOT5njx5EkVRtNbDeJCIiAjatm1LnTp1aNeuHZMnT+brr7/WKpOXl8f8+fPx9vamXr16/PPPP8TFxfHNN9/QqlUr6tatS2RkJC+//LLMdFaO6TTF3cSJEwkJCSExMVG9ZfLbb7+xefNmFi9eXKoBCiGEEOLxBQYG0qlTJ3bu3Mmvv/7Kjz/+yIwZM1iyZIk6LKK47j4PBXfGI9erV4/U1FR1m76+vtaCa/Xr18fKyorU1FSaNWtGamoq77zzjladPj4+hXqWvby8ShTXXUFBQXzxxReMGzcORVFYuXIlI0aMAODatWukpaUxcOBAQkND1WNu375dZFIOoChKsdvevn07U6dO5c8//yQ7O5vbt29z8+ZNrl+/jomJCQAGBgZ4eHioxxw9epT8/HxcXFy06srNzZXOyXJMpyQ6ODgYV1dX5s6dy9q1awFwdXVl165dRY5DEkIIIUTZMzIyokOHDnTo0IFx48YREhLC+PHjCQ4OVh9suzdhfNjDfE9DUUMniqNPnz5ERUVx8OBBbty4wblz5+jduzeAOj558eLFhXIWPT29IutzdnZGo9E88uHB06dP89prrzF48GBiYmKoXLkyu3btYuDAgdy6dUtNoo2NjbVmPMvJyUFPT48DBw4UisHMzKxkJy+eGp2SaIDmzZuzYsWK0oxFCCGEEE+Rm5sb69evB6BatWoAZGRk4OnpCaD1kOG9fv31V2rXrg3A1atXOX78OK6urur+27dvs3//fpo1awbcGcOcmZmplnF1dSUpKYl+/fqpxyQlJeHm5vbQeA0MDIA7M1o8TK1atfD19WXFihXcuHGDDh06YG1tDUD16tWpUaMGp06dIigo6KH13FW5cmX8/Pz47LPPCA8PL5TcZ2ZmYmVlxYEDBygoKCA2Nlb9o+T+oRxF8fT0JD8/n0uXLtGqVatixSTKnk5J9NmzZx+6/+7/WEIIIYQoe1euXKFnz54MGDAADw8PzM3N2b9/PzNmzKBr167And7RFi1aMG3aNBwcHLh06RIffvhhkfVNmjSJKlWqUL16dcaOHUvVqlXp1q2bur9ixYq89957zJ07F319fcLCwmjRooWaVI8aNYpevXrh6elJ+/bt+f7771m7du0jJyewt7dHo9GwceNGOnbsiLGx8QN7aoOCghg/fjy3bt1i9uzZWvsmTpxIeHg4lpaW+Pv7k5uby/79+7l69ao67ON+n332GT4+PjRr1oxJkybh4eHB7du32bZtGwsWLCA1NRUnJyfy8vL49NNP6dy5M0lJSSxcuPCh5wTg4uJCUFAQffv2JTY2Fk9PTy5fvkxCQgIeHh506tTpkXWIMqDoQKPRKBUqVHjgSwihKFlZWQqgZGVllXUoQoj/78aNG0pKSopy48aNsg7lqbp586YyevRopXHjxoqlpaViYmKi1KtXT/nwww+V69evq+VSUlKUli1bKsbGxkqjRo2UrVu3KoCyY8cORVEUZceOHQqgfP/990qDBg0UAwMDpVmzZsrhw4fVOuLi4hRLS0tlzZo1iqOjo2JoaKi0b99eOXPmjFZM8+fPVxwdHZWKFSsqLi4uyrJly7T2A8q6desKncukSZMUGxsbRaPRKP369VMURVF8fX2VYcOGaZW7evWqYmhoqJiYmCj//fdfoXpWrFihNGrUSDEwMFAqVaqktG7dWlm7du1Dr+OFCxeUoUOHKvb29oqBgYFSs2ZNpUuXLur1URRFmTVrlmJra6sYGxsrfn5+yrJlyxRAuXr1qtb1ud+tW7eU6OhopU6dOkrFihUVW1tbpXv37sqRI0ceGpPQzcN+FxT3+1ujKCUYLf//HT58WOtzXl4ehw4dYtasWcTExNCjR49SSO+FeLZlZ2djaWnJ6NGjMTQ0fGjZCRMmPJ2ghHjB3bx5k/T0dBwcHDAyMirrcJ45D1ox8F7x8fFERESQmZn5VGMToiQe9rvg7vd3VlbWQxdL02k4R8OGDQtt8/LyokaNGnz88ceSRAshhBBCiOdaqa4xWa9ePfbt21eaVQohhBBCCFHu6NQTnZ2drfVZURQyMjKYMGECzs7OpRKYEEIIIcqXNm3aPHLO5ODg4BLPOy3Es0inJNrKykprfkO4k0jb2dmxatWqUglMCCGEEEKI8kqnJPr+te0rVKhAtWrVcHJyQl9f56mnhRBCCCGEeCbolPFqNBq8vb0LJcy3b9/ml19+oXXr1qUSnBBCCCGEEOWRTg8Wtm3bln///bfQ9qysLNq2bfvYQQkhhBBCCFGe6ZREK4pSaEw03FkRSdd17oUQQgghhHhWlGg4x935nzUaDcHBwVoLSOTn53PkyBG8vb1LN0IhhBBCCCHKmRL1RFtaWmJpaYmiKJibm6ufLS0tsbGx4Z133mH58uVPKlYhhBBClJH4+PgHrlIoxIuoRD3RcXFxAFSrVo0JEyZgYmICwOnTp1m/fj2urq5UrVq19KMUQgghyrnUmJ+eanuuY9vpdNy5c+cYP348mzdv5p9//sHW1pZu3boRHR1NlSpVAKhTpw4RERFERESUYsRCPF90GhN96NAhli1bBkBmZiYtWrQgNjaWbt26sWDBglINsLQkJiai0WjIzMx8aLk6deowZ86cpxJTaZgwYQKNGjUq6zCeO6dPn0aj0ZCcnFzWoQghRKk5deoUXl5enDhxgpUrV3Ly5EkWLlxIQkICLVu2LHLSgCctLy/vqbcpRGnQOYlu1aoVAN9++y3Vq1fnzJkzLFu2jLlz5xa7nuDgYLp161Zoe3ET3sdRmrelLl++zODBg6lduzaGhobY2Njg5+dHUlJSqdT/MJGRkSQkJDzxdh6kTp06aDQaNBoNJiYmuLu7s2TJkjKLRwghxIMNHToUAwMDtm7diq+vL7Vr1yYgIIDt27fz119/MXbsWNq0acOZM2cYPny4+vv9Xlu2bMHV1RUzMzP8/f3JyMjQ2r9kyRJcXV0xMjKifv36zJ8/X913t4Ni9erV+Pr6YmRkxIoVK57KuQtR2nSaJ/r69euYm5sDsHXrVnr06EGFChVo0aIFZ86cKdUAnwWBgYHcunWLpUuX4ujoyN9//01CQgJXrlzRuc5bt25hYGDwyHJmZmaYmZnp3E5pmDRpEqGhoVy/fp1vvvmG0NBQatasSUBAQJnG9SCKopCfny8LAwkhXij//vsvW7ZsISYmBmNjY619NjY2BAUFsXr1ak6cOEGjRo145513CA0N1Sp3/fp1Zs6cyZdffkmFChV46623iIyMVBPhFStWEB0dzbx58/D09OTQoUOEhoZiampKv3791HpGjx5NbGwsnp6eGBkZPfmTF+IJ0Kkn2snJifXr13Pu3Dm2bNnCq6++CsClS5ewsLAo1QDv2rVrF61atcLY2Bg7OzvCw8O5du2auv/LL7/Ey8sLc3NzbGxsePPNN7l06VKRdSUmJtK/f3+ysrLUv7InTJig7r9+/ToDBgzA3Nyc2rVr87///e+BcWVmZrJz506mT59O27Ztsbe3p1mzZowZM4YuXbpolQsJCaFatWpYWFjQrl07Dh8+rO6/OyxjyZIlODg4YGRkxP/+9z9q1KhBQUGBVptdu3ZlwIABWsfd64svvqBBgwYYGhpia2tLWFhYseM4fPgwbdu2xdzcHAsLC5o0acL+/fsfeP6Aes0dHR2JioqicuXKbNu2rdTafNyf/d07Gz/++CNNmjTB0NCQXbt2UVBQwIwZM3BycsLQ0JDatWsTExOjdW6nTp2ibdu2mJiY0LBhQ/bs2fPA65Cbm0t2drbWSwghyosTJ06gKAqurq5F7nd1deXq1avk5+ejp6en/k61sbFRy+Tl5bFw4UK8vLxo3LgxYWFhWndDx48fT2xsLD169MDBwYEePXowfPhwFi1apNVWRESEWsbW1vbJnLAQT5hOSXR0dDSRkZHUqVOH5s2b07JlS+BOr7Snp2epBgiQlpaGv78/gYGBHDlyhNWrV7Nr1y6t5DAvL4+PPvqIw4cPs379ek6fPk1wcHCR9Xl7ezNnzhwsLCzIyMggIyODyMhIdX9sbCxeXl4cOnSIIUOGMHjwYI4dO1ZkXXd7gtevX09ubu4Dz6Fnz55cunSJH3/8kQMHDtC4cWNeeeUVrfFnJ0+eZM2aNaxdu5bk5GR69uzJlStXtJZZ//fff9m8eTNBQUFFtrNgwQKGDh3KO++8w9GjR9mwYQNOTk7FjiMoKIhatWqxb98+Dhw4wOjRo6lYseIDz+teBQUFrFmzhqtXr2r1oj9Om6X5sx89ejTTpk0jNTUVDw8PxowZw7Rp0xg3bhwpKSl89dVXVK9eXeuYsWPHEhkZSXJyMi4uLvTp04fbt28Xef5Tp07VmrHGzs6uWNdNCCGeJkVRdD7WxMSEunXrqp9tbW3VTotr166RlpbGwIED1e9GMzMzJk+eTFpamlY9Xl5eOscgRHmh0/3s119/nZdffpmMjAwaNmyobn/llVfo3r17ierauHFjoeEI+fn5Wp+nTp1KUFCQ+pSws7Mzc+fOxdfXlwULFmBkZKT2zAI4Ojoyd+5cmjZtSk5OTqH6DQwMsLS0RKPRaP2FfVfHjh0ZMmQIAFFRUcyePZsdO3ZQr169QmX19fWJj48nNDSUhQsX0rhxY3x9ff9fe/ceV1O6/wH8s7vsLrorJVN2JIQiucQMUiYGI9fQHIUYl9wlMV0MMzqGX27H/AxDx5xMMw3i/JLfOJEZhlGNEjUZlGaOLnOim0tUz+8Pr9bP1kWbksvn/Xrt16u9n2et57vW0372d6/9rLUwefJkODg4AHh0JPX8+fMoLCyUrq29ceNGxMbG4rvvvsPs2bMBPJrCsW/fPpiZmUnrHzFiBPbv3w83NzcAj+agm5qa1ntnyHXr1mHZsmVYtGiR9FqfPn0aHUdubi4CAgLQpUsXaV8/TWBgID766CNUVFSgsrISJiYm8PPza5I2m7LvP/74YwwbNgwAUFZWhi1btmD79u3ST4wdO3bE22+/rbRty5cvx8iRIwEAa9asQbdu3XD16lUp1scFBQVh6dKl0vPS0lIm0kT00rC1tYVMJkNmZmadn9WZmZkwNjZW+gx60pMHVWQymZSUl5eXAwB27dqFfv36KdVTV1dXes4bs9Hr4JmORAOP5k/16tULamr/v4q+ffvWmVw0xNXVFampqUqPJ09MS0tLQ2RkpNI3Ww8PD1RXVyM7OxsAkJKSgtGjR8Pa2hr6+voYPHgwACA3N1flbatJfgFIiXZ9U0OAR3Oib968iSNHjmD48OFITEyEk5MTIiMjpfjLy8vRunVrpW3Izs5W+nbevn37WoOXt7c3Dhw4IB3ljoqKwuTJk5X2e43CwkLcvHlTSrif1Jg4li5dCj8/P7i7uyM8PLzW0YO6BAQEIDU1FSdOnEC/fv0QEREhHf1+3jabsu8fP/KRmZmJioqKevdVjcf/F2p+cqzvf0FLSwsGBgZKDyKil0Xr1q0xbNgw7NixA/fu3VMqy8/PR1RUFLy8vCCTySCXy2sd0Hoac3NzWFpa4vr167C1tVV62NjYNOWmEL0UWvzMqlatWilNNwCAP/74Q+l5eXk5PvzwQyxcuLDW8tbW1rhz5w48PDzg4eGBqKgomJmZITc3Fx4eHnjw4IHKMdX1TfvJeclP0tbWxrBhwzBs2DAEBwfDz88PoaGh8PX1RXl5Odq2bYvExMRayz1+hZC6vpmPHj0aQgjExcWhT58++PHHHxEREVFnDE+eKPKkxsQRFhaGqVOnIi4uDvHx8QgNDUV0dHSDvzCYmppKA2VMTAx69OgBZ2dn2NvbP3ebTdn3j+/fp+2rGo//L9Scof60/wUiopfV9u3bMWDAAHh4eGDdunWwsbHB5cuXERAQgHbt2knnhSgUCvzwww+YPHkytLS0Gn0PiDVr1mDhwoUwNDTE8OHDUVFRgeTkZNy+fVvplzqi10GLJ9GN4eTkhIyMjFrJdo309HQUFRUhPDxc+vn8aSfDPcu3bFXY29sjNjYWwKP48/PzoaGhAYVCodJ6tLW1MW7cOERFReHq1avo3LkznJyc6qyrr68PhUKBhISEOqd7NDYOOzs72NnZYcmSJZgyZQr27t3b6Gk6VlZW8PLyQlBQEA4fPvzcbTZH3wOPpoXo6OggISFBmnpCRPQ8nvXmJy9Sp06dkJycjNDQUEyaNAm3bt2ChYUFPD09ERoaChMTEwCPpr99+OGH6NixIyoqKho9j9rPzw+6urr47LPPEBAQgFatWqFHjx68aQu9ll6JJDowMBD9+/eHv78//Pz80KpVK2RkZOD48ePYvn07rK2tIZfLsW3bNsyZMweXLl3C2rVrG1ynQqFAeXk5EhIS4OjoCF1dXekOjKooKirCxIkTMWPGDDg4OEBfXx/JycnYsGEDxowZAwBwd3eHi4sLPD09sWHDBtjZ2eHmzZuIi4vD2LFjn3qChbe3N0aNGoXLly/jgw8+aLBuWFgY5syZgzZt2mDEiBEoKyvDmTNnsGDBgqfG0a1bNwQEBGDChAmwsbHBH3/8gaSkJIwfP16lfbJo0SJ0794dycnJz91mc/Q98OjLSWBgIFasWAG5XI6BAwfizz//xOXLlzFz5kyVtpeI6FXSvn17abphffr37690FSXg0b0dnjxp29PTs1aCPXXqVEydOrXO9SoUiuc6sZHoZfLMc6JfJAcHB5w6dQpXrlzBO++8g169eiEkJASWlpYAHt2GPDIyEjExMbC3t0d4eDg2btzY4DoHDBiAOXPmwMvLC2ZmZtiwYcMzxaanpyfNAx40aBC6d++O4OBgzJo1C9u3bwfwaBrA0aNHMWjQIEyfPh12dnaYPHkybty4UetqEHUZOnQoTExMkJWVVe/AVMPHxwebN2/Gjh070K1bN4waNQq//fZbo+JQV1dHUVERpk2bBjs7O0yaNAkjRozAmjVrVNon9vb2ePfddxESEvLcbTZH39cIDg7GsmXLEBISgq5du8LLy6vBue9ERERENWSCXwmJmkVpaSkMDQ2xcuVK6cok9Xn8OuVE1Hzu37+P7Oxs6Xr8RPRmamgsqPn8LikpafAiAa/EkWgiIiIiopcJk2giIiIiIhUxiSYiIiIiUhGTaCIiIiIiFTGJJiIiIiJSEZNoIiIiIiIVMYkmIiIiIlIRk2giIiJqUomJiZDJZCguLm6wnkKhwObNm19ITE0hLCwMPXv2bOkwXjs5OTmQyWRITU1t6VBU8krc9puIiOhl96JvmqRqe76+viguLkZsbKzS64mJiXB1dcXt27dhZGTUZPE9LjIyEosXL35qUt0Yf/75J0JCQhAXF4eCggIYGxvD0dERISEhGDhw4PMH24Dly5djwYIFzdpGQxQKBW7cuAEA0NHRQceOHbFo0SL4+fm1WExvMibRRM0sKCiowTseERFR440fPx4PHjzA3//+d3To0AEFBQVISEhAUVHRM6/zwYMHkMvlT62np6cHPT29Z26nKXz88ceYNWsW7t69i5iYGMyaNQvt2rXDiBEjWjSu+gghUFVVBQ2N1y/l5HQOIiIiUnL69Gm888470NHRgZWVFRYuXIg7d+5I5V999RWcnZ2hr68PCwsLTJ06FYWFhXWuKzExEdOnT0dJSQlkMhlkMpnSUfS7d+9ixowZ0NfXh7W1Nb744ot64youLsaPP/6Iv/71r3B1dUX79u3Rt29fBAUF4f3331eq5+fnBzMzMxgYGGDo0KFIS0uTymumZezevVu67fMXX3wBS0tLVFdXK7U5ZswYzJgxQ2m5x+3ZswfdunWDlpYW2rZtC39//0bHkZaWBldXV+jr68PAwAC9e/dGcnJyvdsPQNrnHTp0QGBgIExMTHD8+PEma/N5+75mKk98fDx69+4NLS0tnD59GtXV1diwYQNsbW2hpaUFa2trfPLJJ0rbdv36dbi6ukJXVxeOjo44e/Zsg/uipTGJJiIiIsm1a9cwfPhwjB8/HhcvXsQ333yD06dPKyWHDx8+xNq1a5GWlobY2Fjk5OTA19e3zvUNGDAAmzdvhoGBAfLy8pCXl4fly5dL5Zs2bYKzszMuXLiAefPmYe7cucjKyqpzXTVHgmNjY1FRUVHvNkycOBGFhYWIj49HSkoKnJyc4Obmhlu3bkl1rl69igMHDuDgwYNITU3FxIkTUVRUhJMnT0p1bt26hWPHjsHb27vOdj7//HPMnz8fs2fPRnp6Oo4cOQJbW9tGx+Ht7Y233noLSUlJSElJwcqVK6GpqVnvdj2uuroaBw4cwO3bt5WOoj9Pm03Z9ytXrkR4eDgyMzPh4OCAoKAghIeHIzg4GBkZGdi/fz/Mzc2Vllm9ejWWL1+O1NRU2NnZYcqUKaisrGzU/mgJMiGEaOkgiF5HpaWlMDQ0RElJCadzEL0k7t+/j+zsbOnoY1N6FeZE/+Mf/6i13VVVVbh//740J9rPzw/q6urYuXOnVOf06dMYPHgw7ty5U+d+S05ORp8+fVBWVgY9Pb1a86zrmxOtUCjwzjvv4KuvvgLw6Kd/CwsLrFmzBnPmzKlzOw4cOIBZs2bh3r17cHJywuDBgzF58mQ4ODhIsY4cORKFhYXQ0tKSlrO1tcWKFSswe/ZshIWF4dNPP8W///1vmJmZSXU8PT3RunVrfPnllwCAL774AmvWrMHvv/8ONTU1hIWFITY2VjoBrl27dpg+fTrWrVtXK87GxGFgYIBt27bBx8enzm19kkKhQF5eHjQ1NVFRUYHKykqYmJjg559/hq2t7XO32ZR9HxsbizFjxgAAysrKYGZmhu3bt9c5fzsnJwc2NjbYvXs3Zs6cCQDIyMhAt27dkJmZiS5dujRq/6iiobGgsZ/fPBJNRET0hnB1dUVqaqrSY/fu3Up10tLSEBkZKR311dPTg4eHB6qrq5GdnQ0ASElJwejRo2FtbQ19fX0MHjwYAJCbm6tyTDXJLwDIZDJYWFjUOzUEeDQn+ubNmzhy5AiGDx+OxMREODk5ITIyUoq/vLwcrVu3VtqG7OxsXLt2TVpP+/btlRJo4NFR2gMHDkhHuaOiojB58mSoqdVOlwoLC3Hz5k24ubnVGWdj4li6dCn8/Pzg7u6O8PBwpfjqExAQgNTUVJw4cQL9+vVDRESEdPT7edtsyr53dnaW/s7MzERFRUW9+6rG4/8Lbdu2lfbzy+r1m+VNREREdWrVqpXSdAMA+OOPP5Sel5eX48MPP8TChQtrLW9tbY07d+7Aw8MDHh4eiIqKgpmZGXJzc+Hh4YEHDx6oHNOT0xdkMlmteclP0tbWxrBhwzBs2DAEBwfDz88PoaGh8PX1RXl5Odq2bYvExMRayz1+9ZFWrVrVKh89ejSEEIiLi0OfPn3w448/IiIios4YdHR0GoyxMXGEhYVh6tSpiIuLQ3x8PEJDQxEdHY2xY8fWu15TU1PY2trC1tYWMTEx6NGjB5ydnWFvb//cbTZl3z++f5+2r2o8/r8gk8kA4Kn/Cy2JSTQRERFJnJyckJGRUSvZrpGeno6ioiKEh4fDysoKAJ56MpxcLkdVVVWTx1rD3t5eunSfk5MT8vPzoaGhAYVCodJ6tLW1MW7cOERFReHq1avo3LkznJyc6qyrr68PhUKBhIQEuLq61ipvbBx2dnaws7PDkiVLMGXKFOzdu7fBJPpxVlZW8PLyQlBQEA4fPvzcbTZH3wNAp06doKOjg4SEhNfqcnyczkFERESSwMBA/PTTT/D390dqaip+++03HD58WDq5zNraGnK5HNu2bcP169dx5MgRrF27tsF1KhQKlJeXIyEhAf/5z39w9+7dZ4qtqKgIQ4cOxT/+8Q9cvHgR2dnZiImJwYYNG6T5t+7u7nBxcYGnpye+//575OTk4KeffsLq1asblfB5e3sjLi4Oe/bsqfeEwhphYWHYtGkTtm7dit9++w2//PILtm3b1qg47t27B39/fyQmJuLGjRs4c+YMkpKS0LVrV5X2yaJFi/DPf/4TycnJz91mc/Q98OjLSWBgIFasWIF9+/bh2rVrOHfunDT3/FXFI9FERERN4EWfWNhcHBwccOrUKaxevRrvvPMOhBDo2LEjvLy8AABmZmaIjIzEqlWrsHXrVjg5OWHjxo1Kl5h70oABAzBnzhx4eXmhqKgIoaGhz7S/9PT0pHnA165dw8OHD2FlZYVZs2Zh1apVAB5NAzh69ChWr16N6dOn488//4SFhQUGDRpU62oQdRk6dChMTEyQlZWFqVOnNljXx8cH9+/fR0REBJYvXw5TU1NMmDChUXGoq6ujqKgI06ZNQ0FBAUxNTTFu3DisWbNGpX1ib2+Pd999FyEhITh69OhztdkcfV8jODgYGhoaCAkJwc2bN9G2bdt6Tx59VfDqHETNhFfnIHr5NOfVOYjo1cGrcxARERERtQAm0UREREREKmISTURERESkIibRREREREQqYhJNRERERKQiJtFERERERCpiEk1EREREpCIm0UREREREKmISTURERESkIibRRERE1KQSExMhk8lQXFzc0qE0i8jISBgZGUnPw8LC0LNnzwaX8fX1haenp/R8yJAhWLx4sfRcoVBg8+bNTRonNS+Nlg6Anl9iYiJcXV1x+/ZtpTf16yIyMhKLFy+WBuOwsDDExsYiNTW13mV8fX1RXFyM2NhYAI8Gq549e0oDlEKhwOLFi5UGMCKi5/FtTN8X2t6kiedVqu/r64u///3vAAANDQ2YmJjAwcEBU6ZMga+vL9TUmu642oABA5CXlwdDQ8MmW+eThgwZglOnTtV6/eHDh9DQaN70xsvLC++9955Ky2zZsgVCiHrLk5KS0KpVK+m5TCbDoUOHlBJvernwSHQz8fX1hUwmg0wmg6amJszNzTFs2DDs2bMH1dXVTdrWixqsarbn8UdlZWWztVnDy8sLV65cUWmZLVu2IDIyst7ypKQkzJ49W3ouk8mkhJuI6HU1fPhw5OXlIScnB/Hx8XB1dcWiRYswatSoJh3P5XI5LCwsIJPJmmyddZk1axby8vKUHs2dQAOAjo4O2rRpo9IyhoaGDR7oMjMzg66u7nNGRi8Sk+hmxMGqaXCwIiJqGlpaWrCwsEC7du3g5OSEVatW4fDhw4iPj1c68FBcXAw/Pz+YmZnBwMAAQ4cORVpaGgDgypUrkMlk+PXXX5XWHRERgY4dOwKoezrHmTNnMGTIEOjq6sLY2BgeHh64ffs2AKC6uhrr16+HjY0NdHR04OjoiO++++6p26OrqwsLCwulR43du3eja9eu0NbWRpcuXbBjxw6lZQMDA2FnZwddXV106NABwcHBePjwoVSelpYGV1dX6Ovrw8DAAL1790ZycjKA2tM5auzcuRNWVlbQ1dXFpEmTUFJSIpU9OZ3jSY9P51AoFACAsWPHQiaTQaFQICcnB2pqalIMNTZv3oz27ds3+QE6ejom0c2Ig9X/exMGq4qKCpSWlio9iIhedkOHDoWjoyMOHjwovTZx4kQUFhYiPj4eKSkpcHJygpubG27dugU7Ozs4OzsjKipKaT1RUVGYOnVqnW2kpqbCzc0N9vb2OHv2LE6fPo3Ro0ejqqoKALB+/Xrs27cP//3f/43Lly9jyZIl+OCDD+qcrtEYUVFRCAkJwSeffILMzEx8+umnCA4OlqazAIC+vj4iIyORkZGBLVu2YNeuXYiIiJDKvb298dZbbyEpKQkpKSlYuXIlNDU1623z6tWr+Pbbb/HPf/4Tx44dw4ULFzBv3rxnij8pKQkAsHfvXuTl5SEpKQkKhQLu7u7Yu3evUt29e/c2+XQcahzu8ReMg9XrO1itX78ehoaG0sPKyuqZ4iEietG6dOmCnJwcAMDp06dx/vx5xMTEwNnZGZ06dcLGjRthZGQkHXDx9vbG119/LS1/5coVpKSkwNvbu871b9iwAc7OztixYwccHR3RrVs3+Pv7w9TUFBUVFfj000+xZ88eeHh4oEOHDvD19cUHH3yAnTt3Nhj3jh07oKenJz2WLVsGAAgNDcWmTZswbtw42NjYYNy4cViyZInS+j766CMMGDAACoUCo0ePxvLly/Htt99K5bm5uXB3d0eXLl3QqVMnTJw4EY6OjvXGcv/+fezbtw89e/bEoEGDsG3bNkRHRyM/P7/hnV8HMzMzAICRkREsLCyk535+fvj6669RUVEBAPjll1+Qnp6O6dOnq9wGPT8m0S2Ag9XrOVgFBQWhpKREevz+++8qx0JE1BKEENKUwLS0NJSXl6N169ZKY352djauXbsGAJg8eTJycnJw7tw5AI8Opjg5OaFLly51rr/m4E5drl69irt372LYsGFK7e3bt09qrz7e3t5ITU2VHkFBQbhz5w6uXbuGmTNnKq1v3bp1Suv75ptvMHDgQFhYWEBPTw8fffQRcnNzpfKlS5fCz88P7u7uCA8Pf2os1tbWaNeunfTcxcUF1dXVyMrKanA5VXh6ekJdXR2HDh0C8OiXWldXV+kXVXqxeHWOFlDfYPW4e/fuKQ1Wy5cvx7lz59C/f/9GDVYTJ06ss+zxwepxDx48QK9evRqM29vbG6tXr5aeGxkZKQ1Ws2bNksoqKyuVTnT85ptvsHXrVly7dg3l5eWorKyEgYGBVF4zWH311Vdwd3fHxIkTpekqdWlosHp8msnz8PT0xPz583Ho0CFMnjz5qYOVlpYWtLS0mqRtIqIXKTMzEzY2NgCA8vJytG3bFomJibXq1Uyts7CwwNChQ7F//370798f+/fvx9y5c+tdv46OTr1l5eXlAIC4uDilcR3AU8dUQ0ND2NraKr1WUFAAANi1axf69eunVKaurg4AOHv2LLy9vbFmzRp4eHjA0NAQ0dHR2LRpk1Q3LCwMU6dORVxcHOLj4xEaGoro6GiMHTu2wZiak1wux7Rp07B3716MGzcO+/fvx5YtW1osnjcdk+gWwMGKgxUR0cvixIkTSE9Px5IlSwAATk5OyM/Ph4aGRoNHOL29vbFixQpMmTIF169fx+TJk+ut6+DggISEBKxZs6ZWmb29PbS0tJCbm4vBgwc/9/aYm5vD0tIS169fr/cX259++gnt27dXOjB048aNWvXs7OxgZ2eHJUuWYMqUKdi7d2+9n0u5ubm4efMmLC0tAQDnzp2DmpoaOnfu/EzboampKU3DfJyfnx+6d++OHTt2oLKyEuPGjXum9dPzYxL9gnGweoSDFRHRi1dRUYH8/HxUVVWhoKAAx44dw/r16zFq1ChMmzYNAODu7g4XFxd4enpiw4YNsLOzw82bNxEXF4exY8fC2dkZADBu3DjMnTsXc+fOhaurqzQe1yUoKAg9evTAvHnzMGfOHMjlcpw8eRITJ06Eqakpli9fjiVLlqC6uhpvv/02SkpKcObMGRgYGMDHx0fl7VyzZg0WLlwIQ0NDDB8+HBUVFUhOTsbt27exdOlSdOrUCbm5uYiOjkafPn0QFxcnTZEAHv0aHBAQgAkTJsDGxgZ//PEHkpKSMH78+Hrb1NbWho+PDzZu3IjS0lIsXLgQkyZNeuZfRxUKBRISEjBw4EBoaWnB2NgYANC1a1f0798fgYGBmDFjRoMHzqh5MYluRhysOFgR0ZtD1ZuftIRjx46hbdu20NDQgLGxMRwdHbF161b4+PhIJ0zLZDIcPXoUq1evxvTp0/Hnn3/CwsICgwYNgrm5ubQufX19jB49Gt9++y327NnTYLt2dnb4/vvvsWrVKvTt2xc6Ojro168fpkyZAgBYu3YtzMzMsH79ely/fh1GRkbSVa2ehZ+fH3R1dfHZZ58hICAArVq1Qo8ePaQbbL3//vtYsmQJ/P39UVFRgZEjRyI4OBhhYWEAHv2SWlRUhGnTpqGgoACmpqYYN25cnQenatja2mLcuHF47733cOvWLYwaNarWlapUsWnTJixduhS7du1Cu3btpHOpAGDmzJn46aefMGPGjGdePzUBQc3Cx8dHABAAhIaGhjAzMxPu7u5iz549oqqqSqluaWmpWLBggbC0tBSamprCyspKeHt7i9zcXKV6kyZNEgDEnj17lF4/efKkACBu374tvZaYmCgGDBggtLS0hJGRkfDw8JDKq6urxebNm0Xnzp2FpqamMDMzEx4eHuLUqVP1bs/gwYPFokWL6i2PiooSPXv2FHK5XBgbG4tBgwaJgwcPSuUBAQGidevWQk9PT3h5eYmIiAhhaGgohBCioqJCTJ48WVhZWQm5XC4sLS2Fv7+/uHfvnhBCiL1790p1hRAiNDRUODo6ih07dghLS0uhra0tJkyYIG7duqW0/8eMGVNv/O3btxcRERHS8yNHjghbW1uhoaEh2rdvr7RtX375pQAgzp8/X+/216WkpEQAECUlJSotR0TN5969eyIjI0MaX4heRR9//LHo0aNHS4fxSmtoLGjs57dMiAbuQUlEWLt2LWJiYnDx4kWVlistLYWhoSFKSkqUTqIkopZz//59ZGdnw8bGBtra2i0dDpFKysvLkZOTAzc3N6xbt07phH5STUNjQWM/v3mJO6J6lJeX49KlS9i+fTsWLFjQ0uEQEdEbzt/fH71798aQIUM4leMlwCSaqB4crIiI6GUSGRmJiooKfPPNN9IVsKjl8MRConpERkYq3Z6diIiIqAaPRBMR0RuHpwMRvdmaYgxgEk1ERG8MTU1NAMDdu3dbOBIiakk1Y0DNmPAsOJ2DiIjeGOrq6jAyMkJhYSEAQFdXFzKZrIWjIqIXRQiBu3fvorCwEEZGRs81t5xJNBERvVFqbspUk0gT0ZvHyMjomW/QVoNJNBERvVFkMhnatm2LNm3a4OHDhy0dDhG9YJqamk1ydRMm0URE9EZSV1fnZcKI6JnxxEIiIiIiIhUxiSYiIiIiUhGTaCIiIiIiFXFONFEzqbmQe2lpaQtHQkRERI1V87n9tBuyMIkmaiZFRUUAACsrqxaOhIiIiFRVVlYGQ0PDesuZRBM1ExMTEwBAbm5ug29CejmUlpbCysoKv//+OwwMDFo6HGoE9tmrhf316nlT+0wIgbKyMlhaWjZYj0k0UTNRU3t0yoGhoeEbNfi86gwMDNhfrxj22auF/fXqeRP7rDEHv3hiIRERERGRiphEExERERGpiEk0UTPR0tJCaGgotLS0WjoUagT216uHffZqYX+9ethnDZOJp12/g4iIiIiIlPBINBERERGRiphEExERERGpiEk0EREREZGKmEQTEREREamISTRRM/jb3/4GhUIBbW1t9OvXD+fPn2/pkF5LP/zwA0aPHg1LS0vIZDLExsYqlQshEBISgrZt20JHRwfu7u747bfflOrcunUL3t7eMDAwgJGREWbOnIny8nKlOhcvXsQ777wDbW1tWFlZYcOGDbViiYmJQZcuXaCtrY0ePXrg6NGjTb69r7r169ejT58+0NfXR5s2beDp6YmsrCylOvfv38f8+fPRunVr6OnpYfz48SgoKFCqk5ubi5EjR0JXVxdt2rRBQEAAKisrleokJibCyckJWlpasLW1RWRkZK14+D59us8//xwODg7SzTZcXFwQHx8vlbO/Xm7h4eGQyWRYvHix9Br7rAkJImpS0dHRQi6Xiz179ojLly+LWbNmCSMjI1FQUNDSob12jh49KlavXi0OHjwoAIhDhw4plYeHhwtDQ0MRGxsr0tLSxPvvvy9sbGzEvXv3pDrDhw8Xjo6O4ty5c+LHH38Utra2YsqUKVJ5SUmJMDc3F97e3uLSpUvi66+/Fjo6OmLnzp1SnTNnzgh1dXWxYcMGkZGRIT766COhqakp0tPTm30fvEo8PDzE3r17xaVLl0Rqaqp47733hLW1tSgvL5fqzJkzR1hZWYmEhASRnJws+vfvLwYMGCCVV1ZWiu7duwt3d3dx4cIFcfToUWFqaiqCgoKkOtevXxe6urpi6dKlIiMjQ2zbtk2oq6uLY8eOSXX4Pm2cI0eOiLi4OHHlyhWRlZUlVq1aJTQ1NcWlS5eEEOyvl9n58+eFQqEQDg4OYtGiRdLr7LOmwySaqIn17dtXzJ8/X3peVVUlLC0txfr161swqtffk0l0dXW1sLCwEJ999pn0WnFxsdDS0hJff/21EEKIjIwMAUAkJSVJdeLj44VMJhP//ve/hRBC7NixQxgbG4uKigqpTmBgoOjcubP0fNKkSWLkyJFK8fTr1098+OGHTbqNr5vCwkIBQJw6dUoI8ah/NDU1RUxMjFQnMzNTABBnz54VQjz64qSmpiby8/OlOp9//rkwMDCQ+mjFihWiW7duSm15eXkJDw8P6Tnfp8/O2NhY7N69m/31EisrKxOdOnUSx48fF4MHD5aSaPZZ0+J0DqIm9ODBA6SkpMDd3V16TU1NDe7u7jh79mwLRvbmyc7ORn5+vlJfGBoaol+/flJfnD17FkZGRnB2dpbquLu7Q01NDT///LNUZ9CgQZDL5VIdDw8PZGVl4fbt21Kdx9upqcM+b1hJSQkAwMTEBACQkpKChw8fKu3LLl26wNraWqnPevToAXNzc6mOh4cHSktLcfnyZalOQ/3B9+mzqaqqQnR0NO7cuQMXFxf210ts/vz5GDlyZK39yj5rWhotHQDR6+Q///kPqqqqlAYfADA3N8evv/7aQlG9mfLz8wGgzr6oKcvPz0ebNm2UyjU0NGBiYqJUx8bGptY6asqMjY2Rn5/fYDtUW3V1NRYvXoyBAweie/fuAB7tT7lcDiMjI6W6T/ZZXfu6pqyhOqWlpbh37x5u377N96kK0tPT4eLigvv370NPTw+HDh2Cvb09UlNT2V8voejoaPzyyy9ISkqqVcb3WNNiEk1ERC/c/PnzcenSJZw+fbqlQ6Gn6Ny5M1JTU1FSUoLvvvsOPj4+OHXqVEuHRXX4/fffsWjRIhw/fhza2totHc5rj9M5iJqQqakp1NXVa53pXFBQAAsLixaK6s1Us78b6gsLCwsUFhYqlVdWVuLWrVtKdepax+Nt1FeHfV43f39//M///A9OnjyJt956S3rdwsICDx48QHFxsVL9J/vsWfvDwMAAOjo6fJ+qSC6Xw9bWFr1798b69evh6OiILVu2sL9eQikpKSgsLISTkxM0NDSgoaGBU6dOYevWrdDQ0IC5uTn7rAkxiSZqQnK5HL1790ZCQoL0WnV1NRISEuDi4tKCkb15bGxsYGFhodQXpaWl+Pnnn6W+cHFxQXFxMVJSUqQ6J06cQHV1Nfr16yfV+eGHH/Dw4UOpzvHjx9G5c2cYGxtLdR5vp6YO+1yZEAL+/v44dOgQTpw4UWuaTO/evaGpqam0L7OyspCbm6vUZ+np6Upffo4fPw4DAwPY29tLdRrqD75Pn091dTUqKirYXy8hNzc3pKenIzU1VXo4OzvD29tb+pt91oRa+sxGotdNdHS00NLSEpGRkSIjI0PMnj1bGBkZKZ3pTE2jrKxMXLhwQVy4cEEAEP/1X/8lLly4IG7cuCGEeHSJOyMjI3H48GFx8eJFMWbMmDovcderVy/x888/i9OnT4tOnTopXeKuuLhYmJubi7/85S/i0qVLIjo6Wujq6ta6xJ2GhobYuHGjyMzMFKGhobzEXR3mzp0rDA0NRWJiosjLy5Med+/elerMmTNHWFtbixMnTojk5GTh4uIiXFxcpPKay2+9++67IjU1VRw7dkyYmZnVefmtgIAAkZmZKf72t7/Vefktvk+fbuXKleLUqVMiOztbXLx4UaxcuVLIZDLx/fffCyHYX6+Cx6/OIQT7rCkxiSZqBtu2bRPW1tZCLpeLvn37inPnzrV0SK+lkydPCgC1Hj4+PkKIR5e5Cw4OFubm5kJLS0u4ubmJrKwspXUUFRWJKVOmCD09PWFgYCCmT58uysrKlOqkpaWJt99+W2hpaYl27dqJ8PDwWrF8++23ws7OTsjlctGtWzcRFxfXbNv9qqqrrwCIvXv3SnXu3bsn5s2bJ4yNjYWurq4YO3asyMvLU1pPTk6OGDFihNDR0RGmpqZi2bJl4uHDh0p1Tp48KXr27Cnkcrno0KGDUhs1+D59uhkzZoj27dsLuVwuzMzMhJubm5RAC8H+ehU8mUSzz5qOTAghWuYYOBERERHRq4lzoomIiIiIVMQkmoiIiIhIRUyiiYiIiIhUxCSaiIiIiEhFTKKJiIiIiFTEJJqIiIiISEVMoomIiIiIVMQkmoiIiIhIRUyiiYiIiIhUxCSaiIjoBcrJyYFMJkNqampLh0JEz4FJNBERERGRiphEExHRG6W6uhobNmyAra0ttLS0YG1tjU8++QQAkJ6ejqFDh0JHRwetW7fG7NmzUV5eLi07ZMgQLF68WGl9np6e8PX1lZ4rFAp8+umnmDFjBvT19WFtbY0vvvhCKrexsQEA9OrVCzKZDEOGDGm2bSWi5sMkmoiI3ihBQUEIDw9HcHAwMjIysH//fpibm+POnTvw8PCAsbExkpKSEBMTg3/961/w9/dXuY1NmzbB2dkZFy5cwLx58zB37lxkZWUBAM6fPw8A+Ne//oW8vDwcPHiwSbePiF4MjZYOgIiI6EUpKyvDli1bsH37dvj4+AAAOnbsiLfffhu7du3C/fv3sW/fPrRq1QoAsH37dowePRp//etfYW5u3uh23nvvPcybNw8AEBgYiIiICJw8eRKdO3eGmZkZAKB169awsLBo4i0koheFR6KJiOiNkZmZiYqKCri5udVZ5ujoKCXQADBw4EBUV1dLR5Eby8HBQfpbJpPBwsIChYWFzx44Eb10mEQTEdEbQ0dH57mWV1NTgxBC6bWHDx/Wqqepqan0XCaTobq6+rnaJqKXC5NoIiJ6Y3Tq1Ak6OjpISEioVda1a1ekpaXhzp070mtnzpyBmpoaOnfuDAAwMzNDXl6eVF5VVYVLly6pFINcLpeWJaJXF5NoIiJ6Y2hrayMwMBArVqzAvn37cO3aNZw7dw5ffvklvL29oa2tDR8fH1y6dAknT57EggUL8Je//EWaDz106FDExcUhLi4Ov/76K+bOnYvi4mKVYmjTpg10dHRw7NgxFBQUoKSkpBm2lIiaG5NoIiJ6owQHB2PZsmUICQlB165d4eXlhcLCQujq6uJ///d/cevWLfTp0wcTJkyAm5sbtm/fLi07Y8YM+Pj4YNq0aRg8eDA6dOgAV1dXldrX0NDA1q1bsXPnTlhaWmLMmDFNvYlE9ALIxJOTu4iIiIiIqEE8Ek1EREREpCIm0UREREREKmISTURERESkIibRREREREQqYhJNRERERKQiJtFERERERCpiEk1EREREpCIm0UREREREKmISTURERESkIibRREREREQqYhJNRERERKSi/wOXkmTmkX4kkgAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "feature_name = 'study_design_info/primary_purpose'\n",
    "print(feature_name)\n",
    "print('非NA数量:', All[feature_name].notna().sum())\n",
    "print('NMI: ', NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes, continuous_target_discrete[All[feature_name].notna()]))\n",
    "sns.countplot(data=All, y=feature_name, hue=feature_name)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "eligibility/gender\n",
      "非NA数量: 62058\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data2/huyaojun/.conda/envs/openmmlab/lib/python3.8/site-packages/sklearn/metrics/cluster/_supervised.py:66: UserWarning: Clustering metrics expects discrete values but received multiclass values for label, and continuous values for target\n",
      "  warnings.warn(msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "NMI:  0.0318119377822484\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "feature_name = 'eligibility/gender'\n",
    "print(feature_name)\n",
    "print('非NA数量:', All[feature_name].notna().sum())\n",
    "print('NMI: ', NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes, continuous_target_discrete[All[feature_name].notna()]))\n",
    "sns.countplot(data=All, y=feature_name, hue=feature_name)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "eligibility/healthy_volunteers\n",
      "非NA数量: 61964\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/data2/huyaojun/.conda/envs/openmmlab/lib/python3.8/site-packages/sklearn/metrics/cluster/_supervised.py:66: UserWarning: Clustering metrics expects discrete values but received binary values for label, and continuous values for target\n",
      "  warnings.warn(msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "NMI:  0.035295475107702035\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "feature_name = 'eligibility/healthy_volunteers'\n",
    "print(feature_name)\n",
    "print('非NA数量:', All[feature_name].notna().sum())\n",
    "print('NMI: ', NMI(All[All[feature_name].notna()][feature_name].astype('category').cat.codes, continuous_target_discrete[All[feature_name].notna()]))\n",
    "sns.countplot(data=All, y=feature_name, hue=feature_name)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "openmmlab",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}