Sample Weight for Causal Estimation.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Sample Weight for Causal Estimation.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "-SrpI1kP3bsM",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 442
        },
        "outputId": "96f2e900-8093-481f-9c1a-a3e98a60c466"
      },
      "source": [
        "!pip install catboost\n",
        "!pip install category_encoders"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: catboost in /usr/local/lib/python3.6/dist-packages (0.23.2)\n",
            "Requirement already satisfied: plotly in /usr/local/lib/python3.6/dist-packages (from catboost) (4.4.1)\n",
            "Requirement already satisfied: pandas>=0.24.0 in /usr/local/lib/python3.6/dist-packages (from catboost) (1.0.4)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from catboost) (1.12.0)\n",
            "Requirement already satisfied: matplotlib in /usr/local/lib/python3.6/dist-packages (from catboost) (3.2.1)\n",
            "Requirement already satisfied: scipy in /usr/local/lib/python3.6/dist-packages (from catboost) (1.4.1)\n",
            "Requirement already satisfied: graphviz in /usr/local/lib/python3.6/dist-packages (from catboost) (0.10.1)\n",
            "Requirement already satisfied: numpy>=1.16.0 in /usr/local/lib/python3.6/dist-packages (from catboost) (1.18.5)\n",
            "Requirement already satisfied: retrying>=1.3.3 in /usr/local/lib/python3.6/dist-packages (from plotly->catboost) (1.3.3)\n",
            "Requirement already satisfied: python-dateutil>=2.6.1 in /usr/local/lib/python3.6/dist-packages (from pandas>=0.24.0->catboost) (2.8.1)\n",
            "Requirement already satisfied: pytz>=2017.2 in /usr/local/lib/python3.6/dist-packages (from pandas>=0.24.0->catboost) (2018.9)\n",
            "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->catboost) (1.2.0)\n",
            "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->catboost) (2.4.7)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.6/dist-packages (from matplotlib->catboost) (0.10.0)\n",
            "Requirement already satisfied: category_encoders in /usr/local/lib/python3.6/dist-packages (2.2.2)\n",
            "Requirement already satisfied: scipy>=1.0.0 in /usr/local/lib/python3.6/dist-packages (from category_encoders) (1.4.1)\n",
            "Requirement already satisfied: statsmodels>=0.9.0 in /usr/local/lib/python3.6/dist-packages (from category_encoders) (0.10.2)\n",
            "Requirement already satisfied: pandas>=0.21.1 in /usr/local/lib/python3.6/dist-packages (from category_encoders) (1.0.4)\n",
            "Requirement already satisfied: numpy>=1.14.0 in /usr/local/lib/python3.6/dist-packages (from category_encoders) (1.18.5)\n",
            "Requirement already satisfied: patsy>=0.5.1 in /usr/local/lib/python3.6/dist-packages (from category_encoders) (0.5.1)\n",
            "Requirement already satisfied: scikit-learn>=0.20.0 in /usr/local/lib/python3.6/dist-packages (from category_encoders) (0.22.2.post1)\n",
            "Requirement already satisfied: pytz>=2017.2 in /usr/local/lib/python3.6/dist-packages (from pandas>=0.21.1->category_encoders) (2018.9)\n",
            "Requirement already satisfied: python-dateutil>=2.6.1 in /usr/local/lib/python3.6/dist-packages (from pandas>=0.21.1->category_encoders) (2.8.1)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from patsy>=0.5.1->category_encoders) (1.12.0)\n",
            "Requirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.6/dist-packages (from scikit-learn>=0.20.0->category_encoders) (0.15.1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cDxDpyVvbUbE",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 71
        },
        "outputId": "4240284e-d5de-4b4c-87d5-e335683958b5"
      },
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "from scipy import stats\n",
        "from scipy.special import expit\n",
        "\n",
        "import statsmodels.api as sm\n",
        "\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import r2_score, mean_squared_error, mean_absolute_error\n",
        "from sklearn.linear_model import LogisticRegressionCV\n",
        "from sklearn.ensemble import RandomForestClassifier\n",
        "\n",
        "from category_encoders import CatBoostEncoder\n",
        "\n",
        "from lightgbm import LGBMClassifier, LGBMRegressor\n",
        "from catboost import CatBoostClassifier, CatBoostRegressor\n",
        "\n",
        "import matplotlib.pyplot as graph\n",
        "import seaborn as sns\n",
        "\n",
        "\n",
        "graph.style.use('fivethirtyeight')\n",
        "\n",
        "\n",
        "def regression_report(model, x, y):\n",
        "    print(f'Rsq = {r2_score(y, model.predict(x)):.4%}')\n",
        "    print(f'MSE = {mean_squared_error(y, model.predict(x)):.4f}')\n",
        "    print()\n",
        "\n",
        "def causal_report(ψ, true_effect):\n",
        "    abs_error = abs(ψ - true_effect)\n",
        "    print(f'Mean Ψ = {ψ.mean():.4f}')\n",
        "    print('Mean Error of Ψ')\n",
        "    print(f'    {abs_error.mean():.4f}')\n",
        "    print()\n",
        "    return abs_error\n",
        "\n",
        "\n",
        "def ω(g, t):\n",
        "    return (t/g) + ((1-t)/(1-g))\n",
        "\n",
        "\n",
        "# def ω(g, t):\n",
        "#     return np.ones_like(g)\n",
        "\n",
        "\n",
        "true_effect = 1.0\n",
        "N = int(10e3)"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/statsmodels/tools/_testing.py:19: FutureWarning: pandas.util.testing is deprecated. Use the functions in the public API at pandas.testing instead.\n",
            "  import pandas.util.testing as tm\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vnXp8kCZcTvu",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def causal_dataset(n=10000, p=25, treatment_effect=0.5, groups=None, is_confounded=True, baseline=100):\n",
        "    \"\"\"\n",
        "    Simulated causal dataset\n",
        "\n",
        "    :param n:\n",
        "    :param p:\n",
        "    :param treatment_effect:\n",
        "    :param groups: int or list of ints\n",
        "    :param is_confounded: Whether selection bias should be functions\n",
        "    :param baseline:\n",
        "    :return: [df, x_cols, ]\n",
        "    \"\"\"\n",
        "    # Setup quick fake data\n",
        "    m = stats.norm().rvs(p)\n",
        "    x = stats.norm().rvs((n, p))\n",
        "\n",
        "    if is_confounded:\n",
        "        # Confounded Trial (Selection Bias)\n",
        "        true_propensity = expit((x[:, :5] @ m[:5]) + 2 * stats.norm().rvs(n))\n",
        "        treatment_labels = true_propensity.round()\n",
        "    else:\n",
        "        # Randomized Controlled Trial\n",
        "        treatment_labels = stats.bernoulli(p=0.5).rvs(n)\n",
        "\n",
        "    y_oracle = x @ m + treatment_effect * treatment_labels + baseline\n",
        "    y = y_oracle + stats.norm().rvs(n)\n",
        "\n",
        "    # Test model\n",
        "    x_cols = list(range(x.shape[1]))\n",
        "    df_sim = pd.DataFrame(x, columns=x_cols)\n",
        "    df_sim['y'] = y\n",
        "    df_sim['treatment'] = treatment_labels\n",
        "\n",
        "    if groups is not None:\n",
        "        if isinstance(groups, list):\n",
        "            for label, n_cat in zip(list('abcdefghijklmnopqrstuvwxyz'), groups):\n",
        "                df_sim[label] = np.random.choice(list(range(n_cat)), size=n)\n",
        "        elif isinstance(groups, int):\n",
        "            df_sim['a'] = np.random.choice(list(range(groups)), size=n)\n",
        "        else:\n",
        "            raise ValueError('groups must be None, a list of ints or int')\n",
        "\n",
        "    return df_sim, 'y', 'treatment', x_cols"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LB-r_tnM3ylA",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 439
        },
        "outputId": "09588381-1426-4fc1-812a-a704f9cbf08a"
      },
      "source": [
        "df, _, _, x_cols = causal_dataset(n=N, treatment_effect=true_effect)\n",
        "\n",
        "display(df)"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "      <th>2</th>\n",
              "      <th>3</th>\n",
              "      <th>4</th>\n",
              "      <th>5</th>\n",
              "      <th>6</th>\n",
              "      <th>7</th>\n",
              "      <th>8</th>\n",
              "      <th>9</th>\n",
              "      <th>10</th>\n",
              "      <th>11</th>\n",
              "      <th>12</th>\n",
              "      <th>13</th>\n",
              "      <th>14</th>\n",
              "      <th>15</th>\n",
              "      <th>16</th>\n",
              "      <th>17</th>\n",
              "      <th>18</th>\n",
              "      <th>19</th>\n",
              "      <th>20</th>\n",
              "      <th>21</th>\n",
              "      <th>22</th>\n",
              "      <th>23</th>\n",
              "      <th>24</th>\n",
              "      <th>y</th>\n",
              "      <th>treatment</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.000208</td>\n",
              "      <td>0.977570</td>\n",
              "      <td>0.876361</td>\n",
              "      <td>-0.090551</td>\n",
              "      <td>0.025503</td>\n",
              "      <td>0.528804</td>\n",
              "      <td>0.027053</td>\n",
              "      <td>1.987367</td>\n",
              "      <td>0.664258</td>\n",
              "      <td>0.680700</td>\n",
              "      <td>-0.284034</td>\n",
              "      <td>1.915374</td>\n",
              "      <td>-0.757952</td>\n",
              "      <td>-0.264212</td>\n",
              "      <td>-1.975945</td>\n",
              "      <td>0.133856</td>\n",
              "      <td>-0.440575</td>\n",
              "      <td>1.213818</td>\n",
              "      <td>-0.141567</td>\n",
              "      <td>-0.531861</td>\n",
              "      <td>1.199078</td>\n",
              "      <td>-0.431804</td>\n",
              "      <td>-0.866688</td>\n",
              "      <td>0.590372</td>\n",
              "      <td>-1.626178</td>\n",
              "      <td>104.084026</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>-0.347471</td>\n",
              "      <td>0.508242</td>\n",
              "      <td>1.486807</td>\n",
              "      <td>-0.422484</td>\n",
              "      <td>0.728795</td>\n",
              "      <td>1.937878</td>\n",
              "      <td>-0.195505</td>\n",
              "      <td>-0.303565</td>\n",
              "      <td>-1.591880</td>\n",
              "      <td>2.182033</td>\n",
              "      <td>0.395850</td>\n",
              "      <td>-1.580348</td>\n",
              "      <td>-1.584548</td>\n",
              "      <td>-0.400088</td>\n",
              "      <td>-0.118563</td>\n",
              "      <td>0.320088</td>\n",
              "      <td>-0.488115</td>\n",
              "      <td>2.139673</td>\n",
              "      <td>-0.271236</td>\n",
              "      <td>1.898419</td>\n",
              "      <td>-1.016896</td>\n",
              "      <td>1.087069</td>\n",
              "      <td>-0.001104</td>\n",
              "      <td>-0.699625</td>\n",
              "      <td>0.015961</td>\n",
              "      <td>98.792505</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.118143</td>\n",
              "      <td>-0.459323</td>\n",
              "      <td>-1.608429</td>\n",
              "      <td>0.218781</td>\n",
              "      <td>0.809773</td>\n",
              "      <td>0.355525</td>\n",
              "      <td>-0.177288</td>\n",
              "      <td>0.029273</td>\n",
              "      <td>-1.123488</td>\n",
              "      <td>1.910630</td>\n",
              "      <td>1.475551</td>\n",
              "      <td>-0.233513</td>\n",
              "      <td>-0.886838</td>\n",
              "      <td>0.275101</td>\n",
              "      <td>0.533527</td>\n",
              "      <td>1.370779</td>\n",
              "      <td>1.161518</td>\n",
              "      <td>0.301205</td>\n",
              "      <td>0.067991</td>\n",
              "      <td>-0.103004</td>\n",
              "      <td>-0.100747</td>\n",
              "      <td>0.537236</td>\n",
              "      <td>0.024088</td>\n",
              "      <td>-1.630069</td>\n",
              "      <td>-0.766982</td>\n",
              "      <td>97.531371</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>-0.508988</td>\n",
              "      <td>0.406647</td>\n",
              "      <td>-1.196317</td>\n",
              "      <td>-1.148918</td>\n",
              "      <td>0.301679</td>\n",
              "      <td>1.214176</td>\n",
              "      <td>0.276361</td>\n",
              "      <td>-0.276619</td>\n",
              "      <td>0.929199</td>\n",
              "      <td>-0.565612</td>\n",
              "      <td>0.642277</td>\n",
              "      <td>-0.970603</td>\n",
              "      <td>-1.406380</td>\n",
              "      <td>0.121600</td>\n",
              "      <td>0.407611</td>\n",
              "      <td>-1.357076</td>\n",
              "      <td>2.089409</td>\n",
              "      <td>1.347018</td>\n",
              "      <td>-1.521784</td>\n",
              "      <td>2.368646</td>\n",
              "      <td>0.667539</td>\n",
              "      <td>1.128301</td>\n",
              "      <td>-0.851948</td>\n",
              "      <td>-0.459729</td>\n",
              "      <td>1.494309</td>\n",
              "      <td>101.659419</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>2.621434</td>\n",
              "      <td>-1.352550</td>\n",
              "      <td>0.876852</td>\n",
              "      <td>0.716934</td>\n",
              "      <td>0.099097</td>\n",
              "      <td>0.733422</td>\n",
              "      <td>-0.912844</td>\n",
              "      <td>0.097112</td>\n",
              "      <td>-0.336001</td>\n",
              "      <td>0.710181</td>\n",
              "      <td>-1.079322</td>\n",
              "      <td>-0.685772</td>\n",
              "      <td>-1.890344</td>\n",
              "      <td>0.774052</td>\n",
              "      <td>0.388934</td>\n",
              "      <td>0.252038</td>\n",
              "      <td>1.741790</td>\n",
              "      <td>0.325316</td>\n",
              "      <td>0.948764</td>\n",
              "      <td>0.153073</td>\n",
              "      <td>0.330714</td>\n",
              "      <td>1.607198</td>\n",
              "      <td>1.089819</td>\n",
              "      <td>-0.217205</td>\n",
              "      <td>0.022012</td>\n",
              "      <td>104.172250</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9995</th>\n",
              "      <td>1.116552</td>\n",
              "      <td>-1.458677</td>\n",
              "      <td>0.394418</td>\n",
              "      <td>-0.211867</td>\n",
              "      <td>-0.791122</td>\n",
              "      <td>0.709761</td>\n",
              "      <td>-0.597026</td>\n",
              "      <td>-1.129047</td>\n",
              "      <td>-0.584663</td>\n",
              "      <td>-1.182514</td>\n",
              "      <td>1.820127</td>\n",
              "      <td>0.744887</td>\n",
              "      <td>-0.403713</td>\n",
              "      <td>1.104924</td>\n",
              "      <td>-0.026757</td>\n",
              "      <td>-1.590164</td>\n",
              "      <td>0.048047</td>\n",
              "      <td>1.971290</td>\n",
              "      <td>0.192292</td>\n",
              "      <td>0.058989</td>\n",
              "      <td>1.044485</td>\n",
              "      <td>-0.347406</td>\n",
              "      <td>0.814188</td>\n",
              "      <td>0.142175</td>\n",
              "      <td>-1.135331</td>\n",
              "      <td>97.243712</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9996</th>\n",
              "      <td>0.859368</td>\n",
              "      <td>-0.139972</td>\n",
              "      <td>0.276840</td>\n",
              "      <td>1.622946</td>\n",
              "      <td>-0.311682</td>\n",
              "      <td>-0.238094</td>\n",
              "      <td>-1.842512</td>\n",
              "      <td>-1.222325</td>\n",
              "      <td>-0.596717</td>\n",
              "      <td>0.342837</td>\n",
              "      <td>0.146925</td>\n",
              "      <td>2.307062</td>\n",
              "      <td>0.323321</td>\n",
              "      <td>0.236092</td>\n",
              "      <td>0.658827</td>\n",
              "      <td>-0.658557</td>\n",
              "      <td>0.648457</td>\n",
              "      <td>-1.050280</td>\n",
              "      <td>-0.176353</td>\n",
              "      <td>-0.948535</td>\n",
              "      <td>-0.755443</td>\n",
              "      <td>1.235961</td>\n",
              "      <td>0.166930</td>\n",
              "      <td>-0.857952</td>\n",
              "      <td>0.372419</td>\n",
              "      <td>107.093759</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9997</th>\n",
              "      <td>0.250893</td>\n",
              "      <td>1.578939</td>\n",
              "      <td>0.478961</td>\n",
              "      <td>-0.267153</td>\n",
              "      <td>-0.499919</td>\n",
              "      <td>0.976174</td>\n",
              "      <td>-0.582990</td>\n",
              "      <td>0.007184</td>\n",
              "      <td>-1.453816</td>\n",
              "      <td>0.782037</td>\n",
              "      <td>0.313201</td>\n",
              "      <td>0.483566</td>\n",
              "      <td>0.928695</td>\n",
              "      <td>-2.774500</td>\n",
              "      <td>0.245877</td>\n",
              "      <td>-1.627149</td>\n",
              "      <td>0.854330</td>\n",
              "      <td>1.423252</td>\n",
              "      <td>1.071477</td>\n",
              "      <td>-0.122783</td>\n",
              "      <td>1.355418</td>\n",
              "      <td>-2.114104</td>\n",
              "      <td>0.559416</td>\n",
              "      <td>1.005245</td>\n",
              "      <td>-1.128586</td>\n",
              "      <td>94.110420</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9998</th>\n",
              "      <td>-1.359335</td>\n",
              "      <td>0.987536</td>\n",
              "      <td>1.552620</td>\n",
              "      <td>-0.527714</td>\n",
              "      <td>-1.252423</td>\n",
              "      <td>1.487581</td>\n",
              "      <td>0.570776</td>\n",
              "      <td>0.480136</td>\n",
              "      <td>0.488508</td>\n",
              "      <td>-0.788311</td>\n",
              "      <td>-1.002309</td>\n",
              "      <td>-0.905327</td>\n",
              "      <td>1.256796</td>\n",
              "      <td>-0.301201</td>\n",
              "      <td>-0.493345</td>\n",
              "      <td>0.533037</td>\n",
              "      <td>0.472123</td>\n",
              "      <td>-0.118815</td>\n",
              "      <td>1.325220</td>\n",
              "      <td>1.383318</td>\n",
              "      <td>-1.666090</td>\n",
              "      <td>0.850424</td>\n",
              "      <td>0.584376</td>\n",
              "      <td>-0.333658</td>\n",
              "      <td>-0.632610</td>\n",
              "      <td>100.305372</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9999</th>\n",
              "      <td>1.952583</td>\n",
              "      <td>0.574806</td>\n",
              "      <td>-0.117489</td>\n",
              "      <td>1.003257</td>\n",
              "      <td>1.006536</td>\n",
              "      <td>2.085482</td>\n",
              "      <td>0.108971</td>\n",
              "      <td>1.625808</td>\n",
              "      <td>0.327199</td>\n",
              "      <td>-0.910978</td>\n",
              "      <td>0.521816</td>\n",
              "      <td>-1.511342</td>\n",
              "      <td>-1.067815</td>\n",
              "      <td>-0.349784</td>\n",
              "      <td>-0.761299</td>\n",
              "      <td>0.630073</td>\n",
              "      <td>-0.319932</td>\n",
              "      <td>-0.651035</td>\n",
              "      <td>-1.661181</td>\n",
              "      <td>1.350904</td>\n",
              "      <td>1.113419</td>\n",
              "      <td>-0.636560</td>\n",
              "      <td>0.737150</td>\n",
              "      <td>-0.258471</td>\n",
              "      <td>0.449665</td>\n",
              "      <td>98.588177</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>10000 rows × 27 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "             0         1         2  ...        24           y  treatment\n",
              "0     0.000208  0.977570  0.876361  ... -1.626178  104.084026        1.0\n",
              "1    -0.347471  0.508242  1.486807  ...  0.015961   98.792505        0.0\n",
              "2     0.118143 -0.459323 -1.608429  ... -0.766982   97.531371        1.0\n",
              "3    -0.508988  0.406647 -1.196317  ...  1.494309  101.659419        1.0\n",
              "4     2.621434 -1.352550  0.876852  ...  0.022012  104.172250        1.0\n",
              "...        ...       ...       ...  ...       ...         ...        ...\n",
              "9995  1.116552 -1.458677  0.394418  ... -1.135331   97.243712        0.0\n",
              "9996  0.859368 -0.139972  0.276840  ...  0.372419  107.093759        1.0\n",
              "9997  0.250893  1.578939  0.478961  ... -1.128586   94.110420        1.0\n",
              "9998 -1.359335  0.987536  1.552620  ... -0.632610  100.305372        0.0\n",
              "9999  1.952583  0.574806 -0.117489  ...  0.449665   98.588177        0.0\n",
              "\n",
              "[10000 rows x 27 columns]"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "g8GHqKF-441Q",
        "colab_type": "text"
      },
      "source": [
        "# Preprocessing"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FbSiegpQ44oe",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "bc8ea2cf-dea6-4a3b-d387-b420e371579b"
      },
      "source": [
        "x, y, t = df[x_cols + ['treatment']], df['y'], df['treatment']\n",
        "\n",
        "x0, x1 = x.copy(), x.copy()\n",
        "x0['treatment'] = 0\n",
        "x1['treatment'] = 1\n",
        "\n",
        "print(x.shape, y.shape, t.shape)\n",
        "print(x0.shape, x1.shape)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(10000, 26) (10000,) (10000,)\n",
            "(10000, 26) (10000, 26)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pQKjFgbh5kj-",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "d9085b72-1215-4613-9625-b7ff7a9ae1fe"
      },
      "source": [
        "x_train, x_test, y_train, y_test, t_train, t_test = train_test_split(x, y, t, test_size=0.1)\n",
        "\n",
        "print(x_train.shape, x_test.shape, y_train.shape, y_test.shape)"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(9000, 26) (1000, 26) (9000,) (1000,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ojikscvdCXtY",
        "colab_type": "text"
      },
      "source": [
        "# Compute Basic Propensity"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "x4M_lTtWCXhx",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "outputId": "4bc19c4a-2b1f-4dc0-e505-f17009400ba4"
      },
      "source": [
        "%%time\n",
        "g_model = LogisticRegressionCV().fit(x_train.drop(columns=['treatment']), t_train)  # Perfect Model\n",
        "# g_model = RandomForestClassifier(1000, min_samples_leaf=5).fit(x_train.drop(columns=['treatment']), t_train)\n",
        "\n",
        "g_train, g_test, g = [g_model.predict_proba(xi.drop(columns='treatment'))[:, 1] for xi in (x_train, x_test, x)]\n",
        "print(g_train.shape, g_test.shape, g.shape)"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(9000,) (1000,) (10000,)\n",
            "CPU times: user 592 ms, sys: 348 ms, total: 940 ms\n",
            "Wall time: 493 ms\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_Li8Z1BZTHKQ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 303
        },
        "outputId": "b7513b34-c423-405c-aabe-e4cc20fa3d87"
      },
      "source": [
        "graph.title('Propensity Score')\n",
        "sns.distplot(g, kde=False)\n",
        "graph.show()"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OoDEDy1A4PY3",
        "colab_type": "text"
      },
      "source": [
        "# CatBoost S-learner"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Q5hfJKfz4Hyh",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "catboost_kwargs = dict(\n",
        "    iterations=int(10e3),\n",
        "    learning_rate=0.3,\n",
        "    early_stopping_rounds=25,\n",
        "    use_best_model=True\n",
        ")\n",
        "\n",
        "catboost_model = CatBoostRegressor(**catboost_kwargs)\n",
        "catboost_dr_model = CatBoostRegressor(**catboost_kwargs)"
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Xd7lCo1O6Ml_",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "fd6d7804-c0b2-4fc3-cf8d-8f0eca78993a"
      },
      "source": [
        "%%timeit\n",
        "catboost_model.fit(\n",
        "    x_train, y_train,\n",
        "    sample_weight=ω(g_train, t_train),\n",
        "    eval_set=(x_test, y_test),\n",
        "    verbose=False\n",
        ")"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1 loop, best of 3: 2.93 s per loop\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ECwnyC3d6Wdj",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "1ce83017-72a9-4e5f-b491-472a526a7dab"
      },
      "source": [
        "%%time\n",
        "catboost_model.fit(\n",
        "    x_train, y_train,\n",
        "    eval_set=(x_test, y_test)\n",
        ")\n",
        "\n",
        "catboost_dr_model.fit(\n",
        "    x_train, y_train,\n",
        "    sample_weight=ω(g_train, t_train),\n",
        "    eval_set=(x_test, y_test)\n",
        ")"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0:\tlearn: 5.0690929\ttest: 5.1099133\tbest: 5.1099133 (0)\ttotal: 9.95ms\tremaining: 1m 39s\n",
            "1:\tlearn: 4.6155365\ttest: 4.6875701\tbest: 4.6875701 (1)\ttotal: 20.2ms\tremaining: 1m 40s\n",
            "2:\tlearn: 4.2551537\ttest: 4.3166816\tbest: 4.3166816 (2)\ttotal: 29.9ms\tremaining: 1m 39s\n",
            "3:\tlearn: 3.9790951\ttest: 4.0628661\tbest: 4.0628661 (3)\ttotal: 39.7ms\tremaining: 1m 39s\n",
            "4:\tlearn: 3.7288943\ttest: 3.8216580\tbest: 3.8216580 (4)\ttotal: 49.4ms\tremaining: 1m 38s\n",
            "5:\tlearn: 3.5360235\ttest: 3.6369320\tbest: 3.6369320 (5)\ttotal: 59.1ms\tremaining: 1m 38s\n",
            "6:\tlearn: 3.3638711\ttest: 3.4641422\tbest: 3.4641422 (6)\ttotal: 69.1ms\tremaining: 1m 38s\n",
            "7:\tlearn: 3.1901625\ttest: 3.2961604\tbest: 3.2961604 (7)\ttotal: 78.8ms\tremaining: 1m 38s\n",
            "8:\tlearn: 3.0591838\ttest: 3.1683286\tbest: 3.1683286 (8)\ttotal: 88.2ms\tremaining: 1m 37s\n",
            "9:\tlearn: 2.9351591\ttest: 3.0678285\tbest: 3.0678285 (9)\ttotal: 97.5ms\tremaining: 1m 37s\n",
            "10:\tlearn: 2.8329633\ttest: 2.9734955\tbest: 2.9734955 (10)\ttotal: 108ms\tremaining: 1m 37s\n",
            "11:\tlearn: 2.7361608\ttest: 2.8698798\tbest: 2.8698798 (11)\ttotal: 117ms\tremaining: 1m 37s\n",
            "12:\tlearn: 2.6269944\ttest: 2.7503481\tbest: 2.7503481 (12)\ttotal: 126ms\tremaining: 1m 36s\n",
            "13:\tlearn: 2.5369861\ttest: 2.6564130\tbest: 2.6564130 (13)\ttotal: 139ms\tremaining: 1m 39s\n",
            "14:\tlearn: 2.4532112\ttest: 2.5916046\tbest: 2.5916046 (14)\ttotal: 149ms\tremaining: 1m 39s\n",
            "15:\tlearn: 2.3741463\ttest: 2.5027614\tbest: 2.5027614 (15)\ttotal: 159ms\tremaining: 1m 39s\n",
            "16:\tlearn: 2.3005833\ttest: 2.4395260\tbest: 2.4395260 (16)\ttotal: 168ms\tremaining: 1m 38s\n",
            "17:\tlearn: 2.2235661\ttest: 2.3592889\tbest: 2.3592889 (17)\ttotal: 177ms\tremaining: 1m 38s\n",
            "18:\tlearn: 2.1625138\ttest: 2.3149516\tbest: 2.3149516 (18)\ttotal: 187ms\tremaining: 1m 38s\n",
            "19:\tlearn: 2.0982179\ttest: 2.2556659\tbest: 2.2556659 (19)\ttotal: 196ms\tremaining: 1m 37s\n",
            "20:\tlearn: 2.0445342\ttest: 2.2095102\tbest: 2.2095102 (20)\ttotal: 216ms\tremaining: 1m 42s\n",
            "21:\tlearn: 1.9900888\ttest: 2.1477727\tbest: 2.1477727 (21)\ttotal: 226ms\tremaining: 1m 42s\n",
            "22:\tlearn: 1.9409008\ttest: 2.0880634\tbest: 2.0880634 (22)\ttotal: 236ms\tremaining: 1m 42s\n",
            "23:\tlearn: 1.8976269\ttest: 2.0384080\tbest: 2.0384080 (23)\ttotal: 246ms\tremaining: 1m 42s\n",
            "24:\tlearn: 1.8521981\ttest: 1.9910182\tbest: 1.9910182 (24)\ttotal: 256ms\tremaining: 1m 42s\n",
            "25:\tlearn: 1.8077333\ttest: 1.9536373\tbest: 1.9536373 (25)\ttotal: 266ms\tremaining: 1m 41s\n",
            "26:\tlearn: 1.7645879\ttest: 1.9214855\tbest: 1.9214855 (26)\ttotal: 275ms\tremaining: 1m 41s\n",
            "27:\tlearn: 1.7325039\ttest: 1.8878661\tbest: 1.8878661 (27)\ttotal: 285ms\tremaining: 1m 41s\n",
            "28:\tlearn: 1.6947896\ttest: 1.8472109\tbest: 1.8472109 (28)\ttotal: 296ms\tremaining: 1m 41s\n",
            "29:\tlearn: 1.6637549\ttest: 1.8185008\tbest: 1.8185008 (29)\ttotal: 311ms\tremaining: 1m 43s\n",
            "30:\tlearn: 1.6328806\ttest: 1.7908494\tbest: 1.7908494 (30)\ttotal: 321ms\tremaining: 1m 43s\n",
            "31:\tlearn: 1.6012842\ttest: 1.7643867\tbest: 1.7643867 (31)\ttotal: 331ms\tremaining: 1m 43s\n",
            "32:\tlearn: 1.5732032\ttest: 1.7341939\tbest: 1.7341939 (32)\ttotal: 346ms\tremaining: 1m 44s\n",
            "33:\tlearn: 1.5457256\ttest: 1.7155996\tbest: 1.7155996 (33)\ttotal: 360ms\tremaining: 1m 45s\n",
            "34:\tlearn: 1.5197898\ttest: 1.6940649\tbest: 1.6940649 (34)\ttotal: 370ms\tremaining: 1m 45s\n",
            "35:\tlearn: 1.4962943\ttest: 1.6626942\tbest: 1.6626942 (35)\ttotal: 380ms\tremaining: 1m 45s\n",
            "36:\tlearn: 1.4718087\ttest: 1.6414647\tbest: 1.6414647 (36)\ttotal: 389ms\tremaining: 1m 44s\n",
            "37:\tlearn: 1.4499809\ttest: 1.6239369\tbest: 1.6239369 (37)\ttotal: 398ms\tremaining: 1m 44s\n",
            "38:\tlearn: 1.4290604\ttest: 1.6132167\tbest: 1.6132167 (38)\ttotal: 408ms\tremaining: 1m 44s\n",
            "39:\tlearn: 1.4096662\ttest: 1.5961622\tbest: 1.5961622 (39)\ttotal: 422ms\tremaining: 1m 45s\n",
            "40:\tlearn: 1.3908826\ttest: 1.5767160\tbest: 1.5767160 (40)\ttotal: 432ms\tremaining: 1m 45s\n",
            "41:\tlearn: 1.3732111\ttest: 1.5594321\tbest: 1.5594321 (41)\ttotal: 443ms\tremaining: 1m 45s\n",
            "42:\tlearn: 1.3564731\ttest: 1.5423504\tbest: 1.5423504 (42)\ttotal: 453ms\tremaining: 1m 44s\n",
            "43:\tlearn: 1.3420646\ttest: 1.5283187\tbest: 1.5283187 (43)\ttotal: 462ms\tremaining: 1m 44s\n",
            "44:\tlearn: 1.3273220\ttest: 1.5193833\tbest: 1.5193833 (44)\ttotal: 472ms\tremaining: 1m 44s\n",
            "45:\tlearn: 1.3137034\ttest: 1.5102015\tbest: 1.5102015 (45)\ttotal: 481ms\tremaining: 1m 44s\n",
            "46:\tlearn: 1.3007456\ttest: 1.4984366\tbest: 1.4984366 (46)\ttotal: 495ms\tremaining: 1m 44s\n",
            "47:\tlearn: 1.2894329\ttest: 1.4831853\tbest: 1.4831853 (47)\ttotal: 504ms\tremaining: 1m 44s\n",
            "48:\tlearn: 1.2773231\ttest: 1.4728656\tbest: 1.4728656 (48)\ttotal: 514ms\tremaining: 1m 44s\n",
            "49:\tlearn: 1.2659244\ttest: 1.4637242\tbest: 1.4637242 (49)\ttotal: 523ms\tremaining: 1m 44s\n",
            "50:\tlearn: 1.2550973\ttest: 1.4564726\tbest: 1.4564726 (50)\ttotal: 533ms\tremaining: 1m 43s\n",
            "51:\tlearn: 1.2446343\ttest: 1.4506525\tbest: 1.4506525 (51)\ttotal: 542ms\tremaining: 1m 43s\n",
            "52:\tlearn: 1.2337461\ttest: 1.4407386\tbest: 1.4407386 (52)\ttotal: 559ms\tremaining: 1m 44s\n",
            "53:\tlearn: 1.2252886\ttest: 1.4303773\tbest: 1.4303773 (53)\ttotal: 568ms\tremaining: 1m 44s\n",
            "54:\tlearn: 1.2168417\ttest: 1.4223576\tbest: 1.4223576 (54)\ttotal: 578ms\tremaining: 1m 44s\n",
            "55:\tlearn: 1.2091438\ttest: 1.4187106\tbest: 1.4187106 (55)\ttotal: 587ms\tremaining: 1m 44s\n",
            "56:\tlearn: 1.2017117\ttest: 1.4141938\tbest: 1.4141938 (56)\ttotal: 596ms\tremaining: 1m 43s\n",
            "57:\tlearn: 1.1945879\ttest: 1.4067488\tbest: 1.4067488 (57)\ttotal: 606ms\tremaining: 1m 43s\n",
            "58:\tlearn: 1.1874654\ttest: 1.3986797\tbest: 1.3986797 (58)\ttotal: 615ms\tremaining: 1m 43s\n",
            "59:\tlearn: 1.1806666\ttest: 1.3946087\tbest: 1.3946087 (59)\ttotal: 628ms\tremaining: 1m 44s\n",
            "60:\tlearn: 1.1709182\ttest: 1.3836864\tbest: 1.3836864 (60)\ttotal: 637ms\tremaining: 1m 43s\n",
            "61:\tlearn: 1.1624543\ttest: 1.3727925\tbest: 1.3727925 (61)\ttotal: 647ms\tremaining: 1m 43s\n",
            "62:\tlearn: 1.1566638\ttest: 1.3694848\tbest: 1.3694848 (62)\ttotal: 656ms\tremaining: 1m 43s\n",
            "63:\tlearn: 1.1509285\ttest: 1.3643377\tbest: 1.3643377 (63)\ttotal: 666ms\tremaining: 1m 43s\n",
            "64:\tlearn: 1.1457035\ttest: 1.3610244\tbest: 1.3610244 (64)\ttotal: 680ms\tremaining: 1m 43s\n",
            "65:\tlearn: 1.1394781\ttest: 1.3564463\tbest: 1.3564463 (65)\ttotal: 690ms\tremaining: 1m 43s\n",
            "66:\tlearn: 1.1339509\ttest: 1.3501847\tbest: 1.3501847 (66)\ttotal: 702ms\tremaining: 1m 44s\n",
            "67:\tlearn: 1.1274647\ttest: 1.3476082\tbest: 1.3476082 (67)\ttotal: 712ms\tremaining: 1m 43s\n",
            "68:\tlearn: 1.1226581\ttest: 1.3436722\tbest: 1.3436722 (68)\ttotal: 721ms\tremaining: 1m 43s\n",
            "69:\tlearn: 1.1181953\ttest: 1.3393645\tbest: 1.3393645 (69)\ttotal: 730ms\tremaining: 1m 43s\n",
            "70:\tlearn: 1.1126803\ttest: 1.3374056\tbest: 1.3374056 (70)\ttotal: 740ms\tremaining: 1m 43s\n",
            "71:\tlearn: 1.1084321\ttest: 1.3336642\tbest: 1.3336642 (71)\ttotal: 753ms\tremaining: 1m 43s\n",
            "72:\tlearn: 1.1036788\ttest: 1.3336578\tbest: 1.3336578 (72)\ttotal: 763ms\tremaining: 1m 43s\n",
            "73:\tlearn: 1.0993500\ttest: 1.3323264\tbest: 1.3323264 (73)\ttotal: 773ms\tremaining: 1m 43s\n",
            "74:\tlearn: 1.0957862\ttest: 1.3288098\tbest: 1.3288098 (74)\ttotal: 782ms\tremaining: 1m 43s\n",
            "75:\tlearn: 1.0921333\ttest: 1.3262253\tbest: 1.3262253 (75)\ttotal: 793ms\tremaining: 1m 43s\n",
            "76:\tlearn: 1.0887033\ttest: 1.3241603\tbest: 1.3241603 (76)\ttotal: 804ms\tremaining: 1m 43s\n",
            "77:\tlearn: 1.0848549\ttest: 1.3215868\tbest: 1.3215868 (77)\ttotal: 813ms\tremaining: 1m 43s\n",
            "78:\tlearn: 1.0800901\ttest: 1.3187817\tbest: 1.3187817 (78)\ttotal: 822ms\tremaining: 1m 43s\n",
            "79:\tlearn: 1.0756618\ttest: 1.3164585\tbest: 1.3164585 (79)\ttotal: 837ms\tremaining: 1m 43s\n",
            "80:\tlearn: 1.0723304\ttest: 1.3127913\tbest: 1.3127913 (80)\ttotal: 847ms\tremaining: 1m 43s\n",
            "81:\tlearn: 1.0688648\ttest: 1.3120523\tbest: 1.3120523 (81)\ttotal: 856ms\tremaining: 1m 43s\n",
            "82:\tlearn: 1.0654446\ttest: 1.3104687\tbest: 1.3104687 (82)\ttotal: 866ms\tremaining: 1m 43s\n",
            "83:\tlearn: 1.0617675\ttest: 1.3099951\tbest: 1.3099951 (83)\ttotal: 875ms\tremaining: 1m 43s\n",
            "84:\tlearn: 1.0585020\ttest: 1.3072736\tbest: 1.3072736 (84)\ttotal: 884ms\tremaining: 1m 43s\n",
            "85:\tlearn: 1.0552309\ttest: 1.3048513\tbest: 1.3048513 (85)\ttotal: 894ms\tremaining: 1m 43s\n",
            "86:\tlearn: 1.0527532\ttest: 1.3044922\tbest: 1.3044922 (86)\ttotal: 904ms\tremaining: 1m 43s\n",
            "87:\tlearn: 1.0502329\ttest: 1.3026836\tbest: 1.3026836 (87)\ttotal: 917ms\tremaining: 1m 43s\n",
            "88:\tlearn: 1.0461371\ttest: 1.2988172\tbest: 1.2988172 (88)\ttotal: 926ms\tremaining: 1m 43s\n",
            "89:\tlearn: 1.0427171\ttest: 1.2977991\tbest: 1.2977991 (89)\ttotal: 936ms\tremaining: 1m 43s\n",
            "90:\tlearn: 1.0397529\ttest: 1.2963871\tbest: 1.2963871 (90)\ttotal: 945ms\tremaining: 1m 42s\n",
            "91:\tlearn: 1.0360581\ttest: 1.2935081\tbest: 1.2935081 (91)\ttotal: 959ms\tremaining: 1m 43s\n",
            "92:\tlearn: 1.0337249\ttest: 1.2913669\tbest: 1.2913669 (92)\ttotal: 969ms\tremaining: 1m 43s\n",
            "93:\tlearn: 1.0315256\ttest: 1.2898000\tbest: 1.2898000 (93)\ttotal: 978ms\tremaining: 1m 43s\n",
            "94:\tlearn: 1.0286133\ttest: 1.2890249\tbest: 1.2890249 (94)\ttotal: 991ms\tremaining: 1m 43s\n",
            "95:\tlearn: 1.0260063\ttest: 1.2886747\tbest: 1.2886747 (95)\ttotal: 1s\tremaining: 1m 43s\n",
            "96:\tlearn: 1.0235921\ttest: 1.2871268\tbest: 1.2871268 (96)\ttotal: 1.01s\tremaining: 1m 43s\n",
            "97:\tlearn: 1.0205492\ttest: 1.2853743\tbest: 1.2853743 (97)\ttotal: 1.02s\tremaining: 1m 43s\n",
            "98:\tlearn: 1.0183327\ttest: 1.2841596\tbest: 1.2841596 (98)\ttotal: 1.03s\tremaining: 1m 43s\n",
            "99:\tlearn: 1.0153421\ttest: 1.2820298\tbest: 1.2820298 (99)\ttotal: 1.04s\tremaining: 1m 43s\n",
            "100:\tlearn: 1.0130531\ttest: 1.2802743\tbest: 1.2802743 (100)\ttotal: 1.05s\tremaining: 1m 43s\n",
            "101:\tlearn: 1.0110433\ttest: 1.2778776\tbest: 1.2778776 (101)\ttotal: 1.06s\tremaining: 1m 43s\n",
            "102:\tlearn: 1.0076088\ttest: 1.2767635\tbest: 1.2767635 (102)\ttotal: 1.07s\tremaining: 1m 43s\n",
            "103:\tlearn: 1.0048237\ttest: 1.2753418\tbest: 1.2753418 (103)\ttotal: 1.08s\tremaining: 1m 43s\n",
            "104:\tlearn: 1.0029283\ttest: 1.2734213\tbest: 1.2734213 (104)\ttotal: 1.09s\tremaining: 1m 43s\n",
            "105:\tlearn: 1.0001455\ttest: 1.2716083\tbest: 1.2716083 (105)\ttotal: 1.1s\tremaining: 1m 42s\n",
            "106:\tlearn: 0.9971375\ttest: 1.2720520\tbest: 1.2716083 (105)\ttotal: 1.11s\tremaining: 1m 42s\n",
            "107:\tlearn: 0.9943588\ttest: 1.2713601\tbest: 1.2713601 (107)\ttotal: 1.12s\tremaining: 1m 42s\n",
            "108:\tlearn: 0.9913741\ttest: 1.2701172\tbest: 1.2701172 (108)\ttotal: 1.13s\tremaining: 1m 42s\n",
            "109:\tlearn: 0.9897642\ttest: 1.2695407\tbest: 1.2695407 (109)\ttotal: 1.14s\tremaining: 1m 42s\n",
            "110:\tlearn: 0.9877858\ttest: 1.2691319\tbest: 1.2691319 (110)\ttotal: 1.15s\tremaining: 1m 42s\n",
            "111:\tlearn: 0.9856720\ttest: 1.2690442\tbest: 1.2690442 (111)\ttotal: 1.17s\tremaining: 1m 42s\n",
            "112:\tlearn: 0.9838050\ttest: 1.2682764\tbest: 1.2682764 (112)\ttotal: 1.17s\tremaining: 1m 42s\n",
            "113:\tlearn: 0.9821004\ttest: 1.2672670\tbest: 1.2672670 (113)\ttotal: 1.18s\tremaining: 1m 42s\n",
            "114:\tlearn: 0.9799494\ttest: 1.2674907\tbest: 1.2672670 (113)\ttotal: 1.19s\tremaining: 1m 42s\n",
            "115:\tlearn: 0.9777488\ttest: 1.2693138\tbest: 1.2672670 (113)\ttotal: 1.2s\tremaining: 1m 42s\n",
            "116:\tlearn: 0.9758268\ttest: 1.2696357\tbest: 1.2672670 (113)\ttotal: 1.21s\tremaining: 1m 42s\n",
            "117:\tlearn: 0.9737671\ttest: 1.2694210\tbest: 1.2672670 (113)\ttotal: 1.22s\tremaining: 1m 42s\n",
            "118:\tlearn: 0.9711996\ttest: 1.2688175\tbest: 1.2672670 (113)\ttotal: 1.23s\tremaining: 1m 42s\n",
            "119:\tlearn: 0.9686462\ttest: 1.2699165\tbest: 1.2672670 (113)\ttotal: 1.25s\tremaining: 1m 42s\n",
            "120:\tlearn: 0.9661440\ttest: 1.2691907\tbest: 1.2672670 (113)\ttotal: 1.25s\tremaining: 1m 42s\n",
            "121:\tlearn: 0.9643797\ttest: 1.2675095\tbest: 1.2672670 (113)\ttotal: 1.26s\tremaining: 1m 42s\n",
            "122:\tlearn: 0.9624112\ttest: 1.2667510\tbest: 1.2667510 (122)\ttotal: 1.27s\tremaining: 1m 42s\n",
            "123:\tlearn: 0.9601520\ttest: 1.2668504\tbest: 1.2667510 (122)\ttotal: 1.28s\tremaining: 1m 42s\n",
            "124:\tlearn: 0.9576887\ttest: 1.2645922\tbest: 1.2645922 (124)\ttotal: 1.29s\tremaining: 1m 42s\n",
            "125:\tlearn: 0.9553105\ttest: 1.2639029\tbest: 1.2639029 (125)\ttotal: 1.3s\tremaining: 1m 42s\n",
            "126:\tlearn: 0.9539354\ttest: 1.2646058\tbest: 1.2639029 (125)\ttotal: 1.31s\tremaining: 1m 42s\n",
            "127:\tlearn: 0.9512494\ttest: 1.2626992\tbest: 1.2626992 (127)\ttotal: 1.32s\tremaining: 1m 42s\n",
            "128:\tlearn: 0.9491811\ttest: 1.2626581\tbest: 1.2626581 (128)\ttotal: 1.34s\tremaining: 1m 42s\n",
            "129:\tlearn: 0.9468797\ttest: 1.2631100\tbest: 1.2626581 (128)\ttotal: 1.35s\tremaining: 1m 42s\n",
            "130:\tlearn: 0.9444134\ttest: 1.2631738\tbest: 1.2626581 (128)\ttotal: 1.37s\tremaining: 1m 42s\n",
            "131:\tlearn: 0.9425793\ttest: 1.2624126\tbest: 1.2624126 (131)\ttotal: 1.38s\tremaining: 1m 42s\n",
            "132:\tlearn: 0.9405806\ttest: 1.2634317\tbest: 1.2624126 (131)\ttotal: 1.39s\tremaining: 1m 42s\n",
            "133:\tlearn: 0.9386818\ttest: 1.2636178\tbest: 1.2624126 (131)\ttotal: 1.4s\tremaining: 1m 42s\n",
            "134:\tlearn: 0.9365667\ttest: 1.2627214\tbest: 1.2624126 (131)\ttotal: 1.4s\tremaining: 1m 42s\n",
            "135:\tlearn: 0.9348719\ttest: 1.2636537\tbest: 1.2624126 (131)\ttotal: 1.41s\tremaining: 1m 42s\n",
            "136:\tlearn: 0.9329845\ttest: 1.2624122\tbest: 1.2624122 (136)\ttotal: 1.42s\tremaining: 1m 42s\n",
            "137:\tlearn: 0.9309531\ttest: 1.2615831\tbest: 1.2615831 (137)\ttotal: 1.43s\tremaining: 1m 42s\n",
            "138:\tlearn: 0.9286905\ttest: 1.2609731\tbest: 1.2609731 (138)\ttotal: 1.45s\tremaining: 1m 42s\n",
            "139:\tlearn: 0.9262827\ttest: 1.2603588\tbest: 1.2603588 (139)\ttotal: 1.46s\tremaining: 1m 42s\n",
            "140:\tlearn: 0.9245950\ttest: 1.2604845\tbest: 1.2603588 (139)\ttotal: 1.47s\tremaining: 1m 42s\n",
            "141:\tlearn: 0.9226014\ttest: 1.2586907\tbest: 1.2586907 (141)\ttotal: 1.48s\tremaining: 1m 42s\n",
            "142:\tlearn: 0.9206810\ttest: 1.2580572\tbest: 1.2580572 (142)\ttotal: 1.49s\tremaining: 1m 42s\n",
            "143:\tlearn: 0.9183594\ttest: 1.2585803\tbest: 1.2580572 (142)\ttotal: 1.5s\tremaining: 1m 42s\n",
            "144:\tlearn: 0.9163426\ttest: 1.2573636\tbest: 1.2573636 (144)\ttotal: 1.51s\tremaining: 1m 42s\n",
            "145:\tlearn: 0.9139581\ttest: 1.2571099\tbest: 1.2571099 (145)\ttotal: 1.52s\tremaining: 1m 42s\n",
            "146:\tlearn: 0.9124283\ttest: 1.2571833\tbest: 1.2571099 (145)\ttotal: 1.53s\tremaining: 1m 42s\n",
            "147:\tlearn: 0.9101178\ttest: 1.2563551\tbest: 1.2563551 (147)\ttotal: 1.54s\tremaining: 1m 42s\n",
            "148:\tlearn: 0.9075634\ttest: 1.2547884\tbest: 1.2547884 (148)\ttotal: 1.55s\tremaining: 1m 42s\n",
            "149:\tlearn: 0.9057620\ttest: 1.2568421\tbest: 1.2547884 (148)\ttotal: 1.56s\tremaining: 1m 42s\n",
            "150:\tlearn: 0.9037605\ttest: 1.2565733\tbest: 1.2547884 (148)\ttotal: 1.57s\tremaining: 1m 42s\n",
            "151:\tlearn: 0.9020579\ttest: 1.2561845\tbest: 1.2547884 (148)\ttotal: 1.58s\tremaining: 1m 42s\n",
            "152:\tlearn: 0.9000812\ttest: 1.2568343\tbest: 1.2547884 (148)\ttotal: 1.59s\tremaining: 1m 42s\n",
            "153:\tlearn: 0.8969234\ttest: 1.2546345\tbest: 1.2546345 (153)\ttotal: 1.6s\tremaining: 1m 42s\n",
            "154:\tlearn: 0.8946675\ttest: 1.2551513\tbest: 1.2546345 (153)\ttotal: 1.61s\tremaining: 1m 42s\n",
            "155:\tlearn: 0.8926724\ttest: 1.2544400\tbest: 1.2544400 (155)\ttotal: 1.62s\tremaining: 1m 42s\n",
            "156:\tlearn: 0.8910381\ttest: 1.2541608\tbest: 1.2541608 (156)\ttotal: 1.63s\tremaining: 1m 42s\n",
            "157:\tlearn: 0.8897662\ttest: 1.2534572\tbest: 1.2534572 (157)\ttotal: 1.64s\tremaining: 1m 42s\n",
            "158:\tlearn: 0.8880663\ttest: 1.2510162\tbest: 1.2510162 (158)\ttotal: 1.65s\tremaining: 1m 42s\n",
            "159:\tlearn: 0.8863073\ttest: 1.2508379\tbest: 1.2508379 (159)\ttotal: 1.66s\tremaining: 1m 42s\n",
            "160:\tlearn: 0.8833054\ttest: 1.2513280\tbest: 1.2508379 (159)\ttotal: 1.68s\tremaining: 1m 42s\n",
            "161:\tlearn: 0.8814603\ttest: 1.2504026\tbest: 1.2504026 (161)\ttotal: 1.69s\tremaining: 1m 42s\n",
            "162:\tlearn: 0.8789031\ttest: 1.2511356\tbest: 1.2504026 (161)\ttotal: 1.7s\tremaining: 1m 42s\n",
            "163:\tlearn: 0.8771761\ttest: 1.2507445\tbest: 1.2504026 (161)\ttotal: 1.7s\tremaining: 1m 42s\n",
            "164:\tlearn: 0.8752064\ttest: 1.2503396\tbest: 1.2503396 (164)\ttotal: 1.71s\tremaining: 1m 42s\n",
            "165:\tlearn: 0.8734645\ttest: 1.2514200\tbest: 1.2503396 (164)\ttotal: 1.72s\tremaining: 1m 42s\n",
            "166:\tlearn: 0.8724484\ttest: 1.2514522\tbest: 1.2503396 (164)\ttotal: 1.73s\tremaining: 1m 42s\n",
            "167:\tlearn: 0.8705033\ttest: 1.2511244\tbest: 1.2503396 (164)\ttotal: 1.74s\tremaining: 1m 41s\n",
            "168:\tlearn: 0.8673014\ttest: 1.2513377\tbest: 1.2503396 (164)\ttotal: 1.75s\tremaining: 1m 41s\n",
            "169:\tlearn: 0.8660424\ttest: 1.2514016\tbest: 1.2503396 (164)\ttotal: 1.76s\tremaining: 1m 41s\n",
            "170:\tlearn: 0.8646618\ttest: 1.2519365\tbest: 1.2503396 (164)\ttotal: 1.77s\tremaining: 1m 41s\n",
            "171:\tlearn: 0.8630171\ttest: 1.2515977\tbest: 1.2503396 (164)\ttotal: 1.78s\tremaining: 1m 41s\n",
            "172:\tlearn: 0.8608195\ttest: 1.2507356\tbest: 1.2503396 (164)\ttotal: 1.79s\tremaining: 1m 41s\n",
            "173:\tlearn: 0.8584988\ttest: 1.2500140\tbest: 1.2500140 (173)\ttotal: 1.8s\tremaining: 1m 41s\n",
            "174:\tlearn: 0.8563727\ttest: 1.2490647\tbest: 1.2490647 (174)\ttotal: 1.81s\tremaining: 1m 41s\n",
            "175:\tlearn: 0.8542810\ttest: 1.2476676\tbest: 1.2476676 (175)\ttotal: 1.82s\tremaining: 1m 41s\n",
            "176:\tlearn: 0.8525072\ttest: 1.2471618\tbest: 1.2471618 (176)\ttotal: 1.83s\tremaining: 1m 41s\n",
            "177:\tlearn: 0.8509761\ttest: 1.2471221\tbest: 1.2471221 (177)\ttotal: 1.84s\tremaining: 1m 41s\n",
            "178:\tlearn: 0.8498197\ttest: 1.2475917\tbest: 1.2471221 (177)\ttotal: 1.85s\tremaining: 1m 41s\n",
            "179:\tlearn: 0.8484229\ttest: 1.2477423\tbest: 1.2471221 (177)\ttotal: 1.86s\tremaining: 1m 41s\n",
            "180:\tlearn: 0.8465106\ttest: 1.2473287\tbest: 1.2471221 (177)\ttotal: 1.87s\tremaining: 1m 41s\n",
            "181:\tlearn: 0.8447864\ttest: 1.2480739\tbest: 1.2471221 (177)\ttotal: 1.88s\tremaining: 1m 41s\n",
            "182:\tlearn: 0.8429267\ttest: 1.2457838\tbest: 1.2457838 (182)\ttotal: 1.89s\tremaining: 1m 41s\n",
            "183:\tlearn: 0.8412001\ttest: 1.2476298\tbest: 1.2457838 (182)\ttotal: 1.9s\tremaining: 1m 41s\n",
            "184:\tlearn: 0.8396300\ttest: 1.2490782\tbest: 1.2457838 (182)\ttotal: 1.91s\tremaining: 1m 41s\n",
            "185:\tlearn: 0.8381361\ttest: 1.2502642\tbest: 1.2457838 (182)\ttotal: 1.92s\tremaining: 1m 41s\n",
            "186:\tlearn: 0.8356283\ttest: 1.2496998\tbest: 1.2457838 (182)\ttotal: 1.93s\tremaining: 1m 41s\n",
            "187:\tlearn: 0.8339889\ttest: 1.2494584\tbest: 1.2457838 (182)\ttotal: 1.94s\tremaining: 1m 41s\n",
            "188:\tlearn: 0.8321484\ttest: 1.2490182\tbest: 1.2457838 (182)\ttotal: 1.95s\tremaining: 1m 41s\n",
            "189:\tlearn: 0.8307048\ttest: 1.2492876\tbest: 1.2457838 (182)\ttotal: 1.96s\tremaining: 1m 41s\n",
            "190:\tlearn: 0.8292377\ttest: 1.2490486\tbest: 1.2457838 (182)\ttotal: 1.97s\tremaining: 1m 41s\n",
            "191:\tlearn: 0.8269587\ttest: 1.2487730\tbest: 1.2457838 (182)\ttotal: 1.98s\tremaining: 1m 41s\n",
            "192:\tlearn: 0.8247344\ttest: 1.2485556\tbest: 1.2457838 (182)\ttotal: 1.99s\tremaining: 1m 41s\n",
            "193:\tlearn: 0.8224995\ttest: 1.2498415\tbest: 1.2457838 (182)\ttotal: 2s\tremaining: 1m 41s\n",
            "194:\tlearn: 0.8211089\ttest: 1.2496753\tbest: 1.2457838 (182)\ttotal: 2.01s\tremaining: 1m 41s\n",
            "195:\tlearn: 0.8196890\ttest: 1.2498189\tbest: 1.2457838 (182)\ttotal: 2.02s\tremaining: 1m 41s\n",
            "196:\tlearn: 0.8178754\ttest: 1.2503167\tbest: 1.2457838 (182)\ttotal: 2.03s\tremaining: 1m 41s\n",
            "197:\tlearn: 0.8162169\ttest: 1.2502257\tbest: 1.2457838 (182)\ttotal: 2.04s\tremaining: 1m 41s\n",
            "198:\tlearn: 0.8141777\ttest: 1.2504738\tbest: 1.2457838 (182)\ttotal: 2.05s\tremaining: 1m 41s\n",
            "199:\tlearn: 0.8122341\ttest: 1.2514638\tbest: 1.2457838 (182)\ttotal: 2.06s\tremaining: 1m 41s\n",
            "200:\tlearn: 0.8103126\ttest: 1.2503984\tbest: 1.2457838 (182)\ttotal: 2.07s\tremaining: 1m 41s\n",
            "201:\tlearn: 0.8081621\ttest: 1.2502612\tbest: 1.2457838 (182)\ttotal: 2.08s\tremaining: 1m 41s\n",
            "202:\tlearn: 0.8067856\ttest: 1.2498993\tbest: 1.2457838 (182)\ttotal: 2.09s\tremaining: 1m 40s\n",
            "203:\tlearn: 0.8048535\ttest: 1.2496424\tbest: 1.2457838 (182)\ttotal: 2.1s\tremaining: 1m 40s\n",
            "204:\tlearn: 0.8031782\ttest: 1.2484955\tbest: 1.2457838 (182)\ttotal: 2.11s\tremaining: 1m 40s\n",
            "205:\tlearn: 0.8016932\ttest: 1.2481244\tbest: 1.2457838 (182)\ttotal: 2.12s\tremaining: 1m 40s\n",
            "206:\tlearn: 0.8001909\ttest: 1.2477082\tbest: 1.2457838 (182)\ttotal: 2.13s\tremaining: 1m 40s\n",
            "207:\tlearn: 0.7986186\ttest: 1.2476496\tbest: 1.2457838 (182)\ttotal: 2.14s\tremaining: 1m 40s\n",
            "Stopped by overfitting detector  (25 iterations wait)\n",
            "\n",
            "bestTest = 1.245783814\n",
            "bestIteration = 182\n",
            "\n",
            "Shrink model to first 183 iterations.\n",
            "0:\tlearn: 5.0677231\ttest: 5.1965021\tbest: 5.1965021 (0)\ttotal: 9.77ms\tremaining: 1m 37s\n",
            "1:\tlearn: 4.5982572\ttest: 4.7349407\tbest: 4.7349407 (1)\ttotal: 20ms\tremaining: 1m 39s\n",
            "2:\tlearn: 4.2751142\ttest: 4.3401088\tbest: 4.3401088 (2)\ttotal: 29.2ms\tremaining: 1m 37s\n",
            "3:\tlearn: 4.0165592\ttest: 4.1278292\tbest: 4.1278292 (3)\ttotal: 43.6ms\tremaining: 1m 48s\n",
            "4:\tlearn: 3.7595320\ttest: 3.8840206\tbest: 3.8840206 (4)\ttotal: 55.3ms\tremaining: 1m 50s\n",
            "5:\tlearn: 3.5768816\ttest: 3.7145041\tbest: 3.7145041 (5)\ttotal: 64.8ms\tremaining: 1m 47s\n",
            "6:\tlearn: 3.4095583\ttest: 3.5606489\tbest: 3.5606489 (6)\ttotal: 74.6ms\tremaining: 1m 46s\n",
            "7:\tlearn: 3.2372917\ttest: 3.3996084\tbest: 3.3996084 (7)\ttotal: 84.2ms\tremaining: 1m 45s\n",
            "8:\tlearn: 3.0882992\ttest: 3.2398451\tbest: 3.2398451 (8)\ttotal: 93.8ms\tremaining: 1m 44s\n",
            "9:\tlearn: 2.9488929\ttest: 3.1284534\tbest: 3.1284534 (9)\ttotal: 108ms\tremaining: 1m 47s\n",
            "10:\tlearn: 2.8373722\ttest: 3.0250910\tbest: 3.0250910 (10)\ttotal: 120ms\tremaining: 1m 49s\n",
            "11:\tlearn: 2.7394525\ttest: 2.9254417\tbest: 2.9254417 (11)\ttotal: 131ms\tremaining: 1m 48s\n",
            "12:\tlearn: 2.6350684\ttest: 2.8129167\tbest: 2.8129167 (12)\ttotal: 143ms\tremaining: 1m 49s\n",
            "13:\tlearn: 2.5337972\ttest: 2.7190851\tbest: 2.7190851 (13)\ttotal: 152ms\tremaining: 1m 48s\n",
            "14:\tlearn: 2.4470528\ttest: 2.6331910\tbest: 2.6331910 (14)\ttotal: 162ms\tremaining: 1m 47s\n",
            "15:\tlearn: 2.3675192\ttest: 2.5550967\tbest: 2.5550967 (15)\ttotal: 172ms\tremaining: 1m 47s\n",
            "16:\tlearn: 2.3015000\ttest: 2.4893008\tbest: 2.4893008 (16)\ttotal: 181ms\tremaining: 1m 46s\n",
            "17:\tlearn: 2.2446555\ttest: 2.4339919\tbest: 2.4339919 (17)\ttotal: 191ms\tremaining: 1m 45s\n",
            "18:\tlearn: 2.1860481\ttest: 2.3890306\tbest: 2.3890306 (18)\ttotal: 200ms\tremaining: 1m 45s\n",
            "19:\tlearn: 2.1225356\ttest: 2.3288633\tbest: 2.3288633 (19)\ttotal: 213ms\tremaining: 1m 46s\n",
            "20:\tlearn: 2.0609990\ttest: 2.2724322\tbest: 2.2724322 (20)\ttotal: 222ms\tremaining: 1m 45s\n",
            "21:\tlearn: 2.0048218\ttest: 2.2246208\tbest: 2.2246208 (21)\ttotal: 232ms\tremaining: 1m 45s\n",
            "22:\tlearn: 1.9564586\ttest: 2.1676406\tbest: 2.1676406 (22)\ttotal: 245ms\tremaining: 1m 46s\n",
            "23:\tlearn: 1.9151657\ttest: 2.1232805\tbest: 2.1232805 (23)\ttotal: 255ms\tremaining: 1m 45s\n",
            "24:\tlearn: 1.8714717\ttest: 2.0806987\tbest: 2.0806987 (24)\ttotal: 265ms\tremaining: 1m 45s\n",
            "25:\tlearn: 1.8233360\ttest: 2.0308340\tbest: 2.0308340 (25)\ttotal: 275ms\tremaining: 1m 45s\n",
            "26:\tlearn: 1.7809492\ttest: 1.9882667\tbest: 1.9882667 (26)\ttotal: 284ms\tremaining: 1m 44s\n",
            "27:\tlearn: 1.7448379\ttest: 1.9554577\tbest: 1.9554577 (27)\ttotal: 293ms\tremaining: 1m 44s\n",
            "28:\tlearn: 1.7109303\ttest: 1.9200207\tbest: 1.9200207 (28)\ttotal: 303ms\tremaining: 1m 44s\n",
            "29:\tlearn: 1.6756818\ttest: 1.8873867\tbest: 1.8873867 (29)\ttotal: 313ms\tremaining: 1m 43s\n",
            "30:\tlearn: 1.6485263\ttest: 1.8575878\tbest: 1.8575878 (30)\ttotal: 323ms\tremaining: 1m 43s\n",
            "31:\tlearn: 1.6168805\ttest: 1.8259836\tbest: 1.8259836 (31)\ttotal: 332ms\tremaining: 1m 43s\n",
            "32:\tlearn: 1.5880785\ttest: 1.8044286\tbest: 1.8044286 (32)\ttotal: 341ms\tremaining: 1m 43s\n",
            "33:\tlearn: 1.5622258\ttest: 1.7787723\tbest: 1.7787723 (33)\ttotal: 351ms\tremaining: 1m 42s\n",
            "34:\tlearn: 1.5350741\ttest: 1.7478401\tbest: 1.7478401 (34)\ttotal: 360ms\tremaining: 1m 42s\n",
            "35:\tlearn: 1.5080188\ttest: 1.7225978\tbest: 1.7225978 (35)\ttotal: 370ms\tremaining: 1m 42s\n",
            "36:\tlearn: 1.4834708\ttest: 1.7038120\tbest: 1.7038120 (36)\ttotal: 380ms\tremaining: 1m 42s\n",
            "37:\tlearn: 1.4598664\ttest: 1.6803877\tbest: 1.6803877 (37)\ttotal: 391ms\tremaining: 1m 42s\n",
            "38:\tlearn: 1.4383810\ttest: 1.6549117\tbest: 1.6549117 (38)\ttotal: 401ms\tremaining: 1m 42s\n",
            "39:\tlearn: 1.4184811\ttest: 1.6405811\tbest: 1.6405811 (39)\ttotal: 415ms\tremaining: 1m 43s\n",
            "40:\tlearn: 1.4000926\ttest: 1.6254187\tbest: 1.6254187 (40)\ttotal: 425ms\tremaining: 1m 43s\n",
            "41:\tlearn: 1.3827262\ttest: 1.6112386\tbest: 1.6112386 (41)\ttotal: 437ms\tremaining: 1m 43s\n",
            "42:\tlearn: 1.3659475\ttest: 1.5943517\tbest: 1.5943517 (42)\ttotal: 449ms\tremaining: 1m 44s\n",
            "43:\tlearn: 1.3501710\ttest: 1.5777309\tbest: 1.5777309 (43)\ttotal: 459ms\tremaining: 1m 43s\n",
            "44:\tlearn: 1.3332968\ttest: 1.5595650\tbest: 1.5595650 (44)\ttotal: 468ms\tremaining: 1m 43s\n",
            "45:\tlearn: 1.3182956\ttest: 1.5475467\tbest: 1.5475467 (45)\ttotal: 477ms\tremaining: 1m 43s\n",
            "46:\tlearn: 1.3049846\ttest: 1.5393842\tbest: 1.5393842 (46)\ttotal: 486ms\tremaining: 1m 42s\n",
            "47:\tlearn: 1.2931746\ttest: 1.5260943\tbest: 1.5260943 (47)\ttotal: 496ms\tremaining: 1m 42s\n",
            "48:\tlearn: 1.2803180\ttest: 1.5130027\tbest: 1.5130027 (48)\ttotal: 505ms\tremaining: 1m 42s\n",
            "49:\tlearn: 1.2682342\ttest: 1.5015993\tbest: 1.5015993 (49)\ttotal: 515ms\tremaining: 1m 42s\n",
            "50:\tlearn: 1.2572210\ttest: 1.4947241\tbest: 1.4947241 (50)\ttotal: 525ms\tremaining: 1m 42s\n",
            "51:\tlearn: 1.2463211\ttest: 1.4854768\tbest: 1.4854768 (51)\ttotal: 534ms\tremaining: 1m 42s\n",
            "52:\tlearn: 1.2370052\ttest: 1.4729218\tbest: 1.4729218 (52)\ttotal: 544ms\tremaining: 1m 42s\n",
            "53:\tlearn: 1.2283748\ttest: 1.4656608\tbest: 1.4656608 (53)\ttotal: 554ms\tremaining: 1m 41s\n",
            "54:\tlearn: 1.2204515\ttest: 1.4572522\tbest: 1.4572522 (54)\ttotal: 563ms\tremaining: 1m 41s\n",
            "55:\tlearn: 1.2122556\ttest: 1.4502607\tbest: 1.4502607 (55)\ttotal: 573ms\tremaining: 1m 41s\n",
            "56:\tlearn: 1.2043877\ttest: 1.4405420\tbest: 1.4405420 (56)\ttotal: 582ms\tremaining: 1m 41s\n",
            "57:\tlearn: 1.1966478\ttest: 1.4338375\tbest: 1.4338375 (57)\ttotal: 592ms\tremaining: 1m 41s\n",
            "58:\tlearn: 1.1889770\ttest: 1.4271232\tbest: 1.4271232 (58)\ttotal: 601ms\tremaining: 1m 41s\n",
            "59:\tlearn: 1.1828733\ttest: 1.4208363\tbest: 1.4208363 (59)\ttotal: 610ms\tremaining: 1m 41s\n",
            "60:\tlearn: 1.1768948\ttest: 1.4139460\tbest: 1.4139460 (60)\ttotal: 622ms\tremaining: 1m 41s\n",
            "61:\tlearn: 1.1707276\ttest: 1.4079195\tbest: 1.4079195 (61)\ttotal: 632ms\tremaining: 1m 41s\n",
            "62:\tlearn: 1.1643928\ttest: 1.4002885\tbest: 1.4002885 (62)\ttotal: 641ms\tremaining: 1m 41s\n",
            "63:\tlearn: 1.1588888\ttest: 1.3972940\tbest: 1.3972940 (63)\ttotal: 656ms\tremaining: 1m 41s\n",
            "64:\tlearn: 1.1527869\ttest: 1.3924485\tbest: 1.3924485 (64)\ttotal: 666ms\tremaining: 1m 41s\n",
            "65:\tlearn: 1.1479296\ttest: 1.3895977\tbest: 1.3895977 (65)\ttotal: 676ms\tremaining: 1m 41s\n",
            "66:\tlearn: 1.1423514\ttest: 1.3854155\tbest: 1.3854155 (66)\ttotal: 686ms\tremaining: 1m 41s\n",
            "67:\tlearn: 1.1325587\ttest: 1.3829042\tbest: 1.3829042 (67)\ttotal: 696ms\tremaining: 1m 41s\n",
            "68:\tlearn: 1.1277416\ttest: 1.3806592\tbest: 1.3806592 (68)\ttotal: 706ms\tremaining: 1m 41s\n",
            "69:\tlearn: 1.1229444\ttest: 1.3773876\tbest: 1.3773876 (69)\ttotal: 715ms\tremaining: 1m 41s\n",
            "70:\tlearn: 1.1177538\ttest: 1.3731098\tbest: 1.3731098 (70)\ttotal: 725ms\tremaining: 1m 41s\n",
            "71:\tlearn: 1.1122955\ttest: 1.3694655\tbest: 1.3694655 (71)\ttotal: 735ms\tremaining: 1m 41s\n",
            "72:\tlearn: 1.1081469\ttest: 1.3683399\tbest: 1.3683399 (72)\ttotal: 744ms\tremaining: 1m 41s\n",
            "73:\tlearn: 1.1039658\ttest: 1.3666446\tbest: 1.3666446 (73)\ttotal: 753ms\tremaining: 1m 41s\n",
            "74:\tlearn: 1.1000002\ttest: 1.3630920\tbest: 1.3630920 (74)\ttotal: 763ms\tremaining: 1m 40s\n",
            "75:\tlearn: 1.0964384\ttest: 1.3609458\tbest: 1.3609458 (75)\ttotal: 772ms\tremaining: 1m 40s\n",
            "76:\tlearn: 1.0932936\ttest: 1.3576434\tbest: 1.3576434 (76)\ttotal: 782ms\tremaining: 1m 40s\n",
            "77:\tlearn: 1.0883720\ttest: 1.3537069\tbest: 1.3537069 (77)\ttotal: 792ms\tremaining: 1m 40s\n",
            "78:\tlearn: 1.0851368\ttest: 1.3507374\tbest: 1.3507374 (78)\ttotal: 801ms\tremaining: 1m 40s\n",
            "79:\tlearn: 1.0816541\ttest: 1.3485646\tbest: 1.3485646 (79)\ttotal: 817ms\tremaining: 1m 41s\n",
            "80:\tlearn: 1.0783867\ttest: 1.3470134\tbest: 1.3470134 (80)\ttotal: 827ms\tremaining: 1m 41s\n",
            "81:\tlearn: 1.0753529\ttest: 1.3440335\tbest: 1.3440335 (81)\ttotal: 839ms\tremaining: 1m 41s\n",
            "82:\tlearn: 1.0720131\ttest: 1.3416668\tbest: 1.3416668 (82)\ttotal: 848ms\tremaining: 1m 41s\n",
            "83:\tlearn: 1.0684403\ttest: 1.3386211\tbest: 1.3386211 (83)\ttotal: 863ms\tremaining: 1m 41s\n",
            "84:\tlearn: 1.0617843\ttest: 1.3293523\tbest: 1.3293523 (84)\ttotal: 873ms\tremaining: 1m 41s\n",
            "85:\tlearn: 1.0591476\ttest: 1.3282926\tbest: 1.3282926 (85)\ttotal: 882ms\tremaining: 1m 41s\n",
            "86:\tlearn: 1.0549294\ttest: 1.3260909\tbest: 1.3260909 (86)\ttotal: 892ms\tremaining: 1m 41s\n",
            "87:\tlearn: 1.0514542\ttest: 1.3241261\tbest: 1.3241261 (87)\ttotal: 901ms\tremaining: 1m 41s\n",
            "88:\tlearn: 1.0482310\ttest: 1.3222711\tbest: 1.3222711 (88)\ttotal: 915ms\tremaining: 1m 41s\n",
            "89:\tlearn: 1.0454864\ttest: 1.3203546\tbest: 1.3203546 (89)\ttotal: 924ms\tremaining: 1m 41s\n",
            "90:\tlearn: 1.0406955\ttest: 1.3188798\tbest: 1.3188798 (90)\ttotal: 934ms\tremaining: 1m 41s\n",
            "91:\tlearn: 1.0372546\ttest: 1.3166456\tbest: 1.3166456 (91)\ttotal: 944ms\tremaining: 1m 41s\n",
            "92:\tlearn: 1.0343258\ttest: 1.3156963\tbest: 1.3156963 (92)\ttotal: 953ms\tremaining: 1m 41s\n",
            "93:\tlearn: 1.0313570\ttest: 1.3140151\tbest: 1.3140151 (93)\ttotal: 963ms\tremaining: 1m 41s\n",
            "94:\tlearn: 1.0282452\ttest: 1.3131988\tbest: 1.3131988 (94)\ttotal: 973ms\tremaining: 1m 41s\n",
            "95:\tlearn: 1.0249536\ttest: 1.3123070\tbest: 1.3123070 (95)\ttotal: 982ms\tremaining: 1m 41s\n",
            "96:\tlearn: 1.0217945\ttest: 1.3105745\tbest: 1.3105745 (96)\ttotal: 991ms\tremaining: 1m 41s\n",
            "97:\tlearn: 1.0193644\ttest: 1.3087600\tbest: 1.3087600 (97)\ttotal: 1s\tremaining: 1m 41s\n",
            "98:\tlearn: 1.0165593\ttest: 1.3077664\tbest: 1.3077664 (98)\ttotal: 1.01s\tremaining: 1m 41s\n",
            "99:\tlearn: 1.0133237\ttest: 1.3090017\tbest: 1.3077664 (98)\ttotal: 1.02s\tremaining: 1m 41s\n",
            "100:\tlearn: 1.0106129\ttest: 1.3076244\tbest: 1.3076244 (100)\ttotal: 1.03s\tremaining: 1m 41s\n",
            "101:\tlearn: 1.0075514\ttest: 1.3052189\tbest: 1.3052189 (101)\ttotal: 1.04s\tremaining: 1m 41s\n",
            "102:\tlearn: 1.0041746\ttest: 1.3032295\tbest: 1.3032295 (102)\ttotal: 1.05s\tremaining: 1m 41s\n",
            "103:\tlearn: 1.0014181\ttest: 1.3014781\tbest: 1.3014781 (103)\ttotal: 1.06s\tremaining: 1m 41s\n",
            "104:\tlearn: 0.9974313\ttest: 1.2997680\tbest: 1.2997680 (104)\ttotal: 1.07s\tremaining: 1m 41s\n",
            "105:\tlearn: 0.9945557\ttest: 1.2985430\tbest: 1.2985430 (105)\ttotal: 1.08s\tremaining: 1m 41s\n",
            "106:\tlearn: 0.9914224\ttest: 1.2980594\tbest: 1.2980594 (106)\ttotal: 1.1s\tremaining: 1m 42s\n",
            "107:\tlearn: 0.9886140\ttest: 1.2968078\tbest: 1.2968078 (107)\ttotal: 1.11s\tremaining: 1m 42s\n",
            "108:\tlearn: 0.9858134\ttest: 1.2971116\tbest: 1.2968078 (107)\ttotal: 1.13s\tremaining: 1m 42s\n",
            "109:\tlearn: 0.9833192\ttest: 1.2975040\tbest: 1.2968078 (107)\ttotal: 1.13s\tremaining: 1m 42s\n",
            "110:\tlearn: 0.9807493\ttest: 1.2961710\tbest: 1.2961710 (110)\ttotal: 1.14s\tremaining: 1m 41s\n",
            "111:\tlearn: 0.9785801\ttest: 1.2922333\tbest: 1.2922333 (111)\ttotal: 1.15s\tremaining: 1m 41s\n",
            "112:\tlearn: 0.9762338\ttest: 1.2914849\tbest: 1.2914849 (112)\ttotal: 1.16s\tremaining: 1m 41s\n",
            "113:\tlearn: 0.9728494\ttest: 1.2904004\tbest: 1.2904004 (113)\ttotal: 1.17s\tremaining: 1m 41s\n",
            "114:\tlearn: 0.9702530\ttest: 1.2875575\tbest: 1.2875575 (114)\ttotal: 1.18s\tremaining: 1m 41s\n",
            "115:\tlearn: 0.9673343\ttest: 1.2882754\tbest: 1.2875575 (114)\ttotal: 1.19s\tremaining: 1m 41s\n",
            "116:\tlearn: 0.9649831\ttest: 1.2882943\tbest: 1.2875575 (114)\ttotal: 1.2s\tremaining: 1m 41s\n",
            "117:\tlearn: 0.9625204\ttest: 1.2876232\tbest: 1.2875575 (114)\ttotal: 1.21s\tremaining: 1m 41s\n",
            "118:\tlearn: 0.9601084\ttest: 1.2864855\tbest: 1.2864855 (118)\ttotal: 1.22s\tremaining: 1m 41s\n",
            "119:\tlearn: 0.9577118\ttest: 1.2870292\tbest: 1.2864855 (118)\ttotal: 1.23s\tremaining: 1m 41s\n",
            "120:\tlearn: 0.9550232\ttest: 1.2859114\tbest: 1.2859114 (120)\ttotal: 1.24s\tremaining: 1m 41s\n",
            "121:\tlearn: 0.9526174\ttest: 1.2860503\tbest: 1.2859114 (120)\ttotal: 1.25s\tremaining: 1m 41s\n",
            "122:\tlearn: 0.9505802\ttest: 1.2857714\tbest: 1.2857714 (122)\ttotal: 1.27s\tremaining: 1m 41s\n",
            "123:\tlearn: 0.9487920\ttest: 1.2869937\tbest: 1.2857714 (122)\ttotal: 1.27s\tremaining: 1m 41s\n",
            "124:\tlearn: 0.9470013\ttest: 1.2871937\tbest: 1.2857714 (122)\ttotal: 1.28s\tremaining: 1m 41s\n",
            "125:\tlearn: 0.9441995\ttest: 1.2877927\tbest: 1.2857714 (122)\ttotal: 1.29s\tremaining: 1m 41s\n",
            "126:\tlearn: 0.9420867\ttest: 1.2865548\tbest: 1.2857714 (122)\ttotal: 1.3s\tremaining: 1m 41s\n",
            "127:\tlearn: 0.9401649\ttest: 1.2862671\tbest: 1.2857714 (122)\ttotal: 1.31s\tremaining: 1m 41s\n",
            "128:\tlearn: 0.9384214\ttest: 1.2855863\tbest: 1.2855863 (128)\ttotal: 1.32s\tremaining: 1m 41s\n",
            "129:\tlearn: 0.9366669\ttest: 1.2857049\tbest: 1.2855863 (128)\ttotal: 1.33s\tremaining: 1m 41s\n",
            "130:\tlearn: 0.9333761\ttest: 1.2816523\tbest: 1.2816523 (130)\ttotal: 1.34s\tremaining: 1m 41s\n",
            "131:\tlearn: 0.9312211\ttest: 1.2816363\tbest: 1.2816363 (131)\ttotal: 1.35s\tremaining: 1m 41s\n",
            "132:\tlearn: 0.9283802\ttest: 1.2803091\tbest: 1.2803091 (132)\ttotal: 1.36s\tremaining: 1m 41s\n",
            "133:\tlearn: 0.9255986\ttest: 1.2791728\tbest: 1.2791728 (133)\ttotal: 1.37s\tremaining: 1m 41s\n",
            "134:\tlearn: 0.9231705\ttest: 1.2798353\tbest: 1.2791728 (133)\ttotal: 1.38s\tremaining: 1m 41s\n",
            "135:\tlearn: 0.9206715\ttest: 1.2792209\tbest: 1.2791728 (133)\ttotal: 1.39s\tremaining: 1m 40s\n",
            "136:\tlearn: 0.9186443\ttest: 1.2774465\tbest: 1.2774465 (136)\ttotal: 1.4s\tremaining: 1m 40s\n",
            "137:\tlearn: 0.9166540\ttest: 1.2759735\tbest: 1.2759735 (137)\ttotal: 1.41s\tremaining: 1m 40s\n",
            "138:\tlearn: 0.9139654\ttest: 1.2748003\tbest: 1.2748003 (138)\ttotal: 1.42s\tremaining: 1m 41s\n",
            "139:\tlearn: 0.9112962\ttest: 1.2760513\tbest: 1.2748003 (138)\ttotal: 1.43s\tremaining: 1m 40s\n",
            "140:\tlearn: 0.9084600\ttest: 1.2752385\tbest: 1.2748003 (138)\ttotal: 1.44s\tremaining: 1m 40s\n",
            "141:\tlearn: 0.9056434\ttest: 1.2748934\tbest: 1.2748003 (138)\ttotal: 1.45s\tremaining: 1m 40s\n",
            "142:\tlearn: 0.9032510\ttest: 1.2729115\tbest: 1.2729115 (142)\ttotal: 1.47s\tremaining: 1m 41s\n",
            "143:\tlearn: 0.9014120\ttest: 1.2732186\tbest: 1.2729115 (142)\ttotal: 1.48s\tremaining: 1m 40s\n",
            "144:\tlearn: 0.8996329\ttest: 1.2728785\tbest: 1.2728785 (144)\ttotal: 1.48s\tremaining: 1m 40s\n",
            "145:\tlearn: 0.8974978\ttest: 1.2718286\tbest: 1.2718286 (145)\ttotal: 1.49s\tremaining: 1m 40s\n",
            "146:\tlearn: 0.8957508\ttest: 1.2725329\tbest: 1.2718286 (145)\ttotal: 1.5s\tremaining: 1m 40s\n",
            "147:\tlearn: 0.8927590\ttest: 1.2729640\tbest: 1.2718286 (145)\ttotal: 1.51s\tremaining: 1m 40s\n",
            "148:\tlearn: 0.8910373\ttest: 1.2728420\tbest: 1.2718286 (145)\ttotal: 1.52s\tremaining: 1m 40s\n",
            "149:\tlearn: 0.8895763\ttest: 1.2724616\tbest: 1.2718286 (145)\ttotal: 1.53s\tremaining: 1m 40s\n",
            "150:\tlearn: 0.8876258\ttest: 1.2727490\tbest: 1.2718286 (145)\ttotal: 1.54s\tremaining: 1m 40s\n",
            "151:\tlearn: 0.8855618\ttest: 1.2726329\tbest: 1.2718286 (145)\ttotal: 1.55s\tremaining: 1m 40s\n",
            "152:\tlearn: 0.8839638\ttest: 1.2704054\tbest: 1.2704054 (152)\ttotal: 1.56s\tremaining: 1m 40s\n",
            "153:\tlearn: 0.8813992\ttest: 1.2703196\tbest: 1.2703196 (153)\ttotal: 1.57s\tremaining: 1m 40s\n",
            "154:\tlearn: 0.8799097\ttest: 1.2707443\tbest: 1.2703196 (153)\ttotal: 1.58s\tremaining: 1m 40s\n",
            "155:\tlearn: 0.8770078\ttest: 1.2723867\tbest: 1.2703196 (153)\ttotal: 1.59s\tremaining: 1m 40s\n",
            "156:\tlearn: 0.8749306\ttest: 1.2718334\tbest: 1.2703196 (153)\ttotal: 1.6s\tremaining: 1m 40s\n",
            "157:\tlearn: 0.8727905\ttest: 1.2706172\tbest: 1.2703196 (153)\ttotal: 1.61s\tremaining: 1m 40s\n",
            "158:\tlearn: 0.8709138\ttest: 1.2700674\tbest: 1.2700674 (158)\ttotal: 1.62s\tremaining: 1m 40s\n",
            "159:\tlearn: 0.8692739\ttest: 1.2697703\tbest: 1.2697703 (159)\ttotal: 1.63s\tremaining: 1m 40s\n",
            "160:\tlearn: 0.8676065\ttest: 1.2691862\tbest: 1.2691862 (160)\ttotal: 1.64s\tremaining: 1m 40s\n",
            "161:\tlearn: 0.8656542\ttest: 1.2664871\tbest: 1.2664871 (161)\ttotal: 1.65s\tremaining: 1m 40s\n",
            "162:\tlearn: 0.8632810\ttest: 1.2672933\tbest: 1.2664871 (161)\ttotal: 1.67s\tremaining: 1m 40s\n",
            "163:\tlearn: 0.8612764\ttest: 1.2666491\tbest: 1.2664871 (161)\ttotal: 1.68s\tremaining: 1m 40s\n",
            "164:\tlearn: 0.8597288\ttest: 1.2670010\tbest: 1.2664871 (161)\ttotal: 1.69s\tremaining: 1m 40s\n",
            "165:\tlearn: 0.8578600\ttest: 1.2677525\tbest: 1.2664871 (161)\ttotal: 1.7s\tremaining: 1m 40s\n",
            "166:\tlearn: 0.8558650\ttest: 1.2668702\tbest: 1.2664871 (161)\ttotal: 1.71s\tremaining: 1m 40s\n",
            "167:\tlearn: 0.8539159\ttest: 1.2657527\tbest: 1.2657527 (167)\ttotal: 1.72s\tremaining: 1m 40s\n",
            "168:\tlearn: 0.8515825\ttest: 1.2664568\tbest: 1.2657527 (167)\ttotal: 1.73s\tremaining: 1m 40s\n",
            "169:\tlearn: 0.8498355\ttest: 1.2657734\tbest: 1.2657527 (167)\ttotal: 1.74s\tremaining: 1m 40s\n",
            "170:\tlearn: 0.8475473\ttest: 1.2659493\tbest: 1.2657527 (167)\ttotal: 1.75s\tremaining: 1m 40s\n",
            "171:\tlearn: 0.8458529\ttest: 1.2652545\tbest: 1.2652545 (171)\ttotal: 1.76s\tremaining: 1m 40s\n",
            "172:\tlearn: 0.8440280\ttest: 1.2651577\tbest: 1.2651577 (172)\ttotal: 1.77s\tremaining: 1m 40s\n",
            "173:\tlearn: 0.8421665\ttest: 1.2651703\tbest: 1.2651577 (172)\ttotal: 1.78s\tremaining: 1m 40s\n",
            "174:\tlearn: 0.8407275\ttest: 1.2640843\tbest: 1.2640843 (174)\ttotal: 1.79s\tremaining: 1m 40s\n",
            "175:\tlearn: 0.8391341\ttest: 1.2637853\tbest: 1.2637853 (175)\ttotal: 1.8s\tremaining: 1m 40s\n",
            "176:\tlearn: 0.8371045\ttest: 1.2645214\tbest: 1.2637853 (175)\ttotal: 1.81s\tremaining: 1m 40s\n",
            "177:\tlearn: 0.8351407\ttest: 1.2630892\tbest: 1.2630892 (177)\ttotal: 1.83s\tremaining: 1m 40s\n",
            "178:\tlearn: 0.8333533\ttest: 1.2629324\tbest: 1.2629324 (178)\ttotal: 1.84s\tremaining: 1m 40s\n",
            "179:\tlearn: 0.8310844\ttest: 1.2629632\tbest: 1.2629324 (178)\ttotal: 1.85s\tremaining: 1m 40s\n",
            "180:\tlearn: 0.8294160\ttest: 1.2632663\tbest: 1.2629324 (178)\ttotal: 1.86s\tremaining: 1m 40s\n",
            "181:\tlearn: 0.8271176\ttest: 1.2635755\tbest: 1.2629324 (178)\ttotal: 1.87s\tremaining: 1m 40s\n",
            "182:\tlearn: 0.8254601\ttest: 1.2615875\tbest: 1.2615875 (182)\ttotal: 1.88s\tremaining: 1m 40s\n",
            "183:\tlearn: 0.8234093\ttest: 1.2621464\tbest: 1.2615875 (182)\ttotal: 1.89s\tremaining: 1m 40s\n",
            "184:\tlearn: 0.8222639\ttest: 1.2612289\tbest: 1.2612289 (184)\ttotal: 1.9s\tremaining: 1m 40s\n",
            "185:\tlearn: 0.8196729\ttest: 1.2594420\tbest: 1.2594420 (185)\ttotal: 1.91s\tremaining: 1m 40s\n",
            "186:\tlearn: 0.8177904\ttest: 1.2595472\tbest: 1.2594420 (185)\ttotal: 1.92s\tremaining: 1m 40s\n",
            "187:\tlearn: 0.8159892\ttest: 1.2592710\tbest: 1.2592710 (187)\ttotal: 1.93s\tremaining: 1m 40s\n",
            "188:\tlearn: 0.8146280\ttest: 1.2590289\tbest: 1.2590289 (188)\ttotal: 1.94s\tremaining: 1m 40s\n",
            "189:\tlearn: 0.8126710\ttest: 1.2603247\tbest: 1.2590289 (188)\ttotal: 1.95s\tremaining: 1m 40s\n",
            "190:\tlearn: 0.8100080\ttest: 1.2593781\tbest: 1.2590289 (188)\ttotal: 1.96s\tremaining: 1m 40s\n",
            "191:\tlearn: 0.8085114\ttest: 1.2599658\tbest: 1.2590289 (188)\ttotal: 1.97s\tremaining: 1m 40s\n",
            "192:\tlearn: 0.8067994\ttest: 1.2606255\tbest: 1.2590289 (188)\ttotal: 1.98s\tremaining: 1m 40s\n",
            "193:\tlearn: 0.8045156\ttest: 1.2610026\tbest: 1.2590289 (188)\ttotal: 1.99s\tremaining: 1m 40s\n",
            "194:\tlearn: 0.8034581\ttest: 1.2611235\tbest: 1.2590289 (188)\ttotal: 2s\tremaining: 1m 40s\n",
            "195:\tlearn: 0.8021855\ttest: 1.2609031\tbest: 1.2590289 (188)\ttotal: 2.01s\tremaining: 1m 40s\n",
            "196:\tlearn: 0.8005507\ttest: 1.2597935\tbest: 1.2590289 (188)\ttotal: 2.03s\tremaining: 1m 40s\n",
            "197:\tlearn: 0.7989294\ttest: 1.2603881\tbest: 1.2590289 (188)\ttotal: 2.04s\tremaining: 1m 40s\n",
            "198:\tlearn: 0.7976095\ttest: 1.2602955\tbest: 1.2590289 (188)\ttotal: 2.05s\tremaining: 1m 40s\n",
            "199:\tlearn: 0.7957582\ttest: 1.2604191\tbest: 1.2590289 (188)\ttotal: 2.06s\tremaining: 1m 40s\n",
            "200:\tlearn: 0.7944222\ttest: 1.2599580\tbest: 1.2590289 (188)\ttotal: 2.07s\tremaining: 1m 41s\n",
            "201:\tlearn: 0.7932035\ttest: 1.2603882\tbest: 1.2590289 (188)\ttotal: 2.09s\tremaining: 1m 41s\n",
            "202:\tlearn: 0.7913294\ttest: 1.2594193\tbest: 1.2590289 (188)\ttotal: 2.1s\tremaining: 1m 41s\n",
            "203:\tlearn: 0.7897803\ttest: 1.2596733\tbest: 1.2590289 (188)\ttotal: 2.11s\tremaining: 1m 41s\n",
            "204:\tlearn: 0.7875487\ttest: 1.2604019\tbest: 1.2590289 (188)\ttotal: 2.12s\tremaining: 1m 41s\n",
            "205:\tlearn: 0.7853104\ttest: 1.2605423\tbest: 1.2590289 (188)\ttotal: 2.13s\tremaining: 1m 41s\n",
            "206:\tlearn: 0.7841861\ttest: 1.2610009\tbest: 1.2590289 (188)\ttotal: 2.14s\tremaining: 1m 41s\n",
            "207:\tlearn: 0.7829719\ttest: 1.2615650\tbest: 1.2590289 (188)\ttotal: 2.15s\tremaining: 1m 41s\n",
            "208:\tlearn: 0.7812576\ttest: 1.2604202\tbest: 1.2590289 (188)\ttotal: 2.16s\tremaining: 1m 41s\n",
            "209:\tlearn: 0.7790086\ttest: 1.2596239\tbest: 1.2590289 (188)\ttotal: 2.17s\tremaining: 1m 41s\n",
            "210:\tlearn: 0.7769224\ttest: 1.2592877\tbest: 1.2590289 (188)\ttotal: 2.18s\tremaining: 1m 41s\n",
            "211:\tlearn: 0.7745308\ttest: 1.2587253\tbest: 1.2587253 (211)\ttotal: 2.19s\tremaining: 1m 40s\n",
            "212:\tlearn: 0.7728382\ttest: 1.2576958\tbest: 1.2576958 (212)\ttotal: 2.2s\tremaining: 1m 40s\n",
            "213:\tlearn: 0.7715263\ttest: 1.2567638\tbest: 1.2567638 (213)\ttotal: 2.21s\tremaining: 1m 40s\n",
            "214:\tlearn: 0.7699025\ttest: 1.2575510\tbest: 1.2567638 (213)\ttotal: 2.22s\tremaining: 1m 40s\n",
            "215:\tlearn: 0.7681521\ttest: 1.2566322\tbest: 1.2566322 (215)\ttotal: 2.23s\tremaining: 1m 41s\n",
            "216:\tlearn: 0.7663877\ttest: 1.2566504\tbest: 1.2566322 (215)\ttotal: 2.24s\tremaining: 1m 41s\n",
            "217:\tlearn: 0.7650969\ttest: 1.2565218\tbest: 1.2565218 (217)\ttotal: 2.25s\tremaining: 1m 40s\n",
            "218:\tlearn: 0.7633583\ttest: 1.2565661\tbest: 1.2565218 (217)\ttotal: 2.26s\tremaining: 1m 40s\n",
            "219:\tlearn: 0.7619712\ttest: 1.2562605\tbest: 1.2562605 (219)\ttotal: 2.27s\tremaining: 1m 40s\n",
            "220:\tlearn: 0.7604907\ttest: 1.2570364\tbest: 1.2562605 (219)\ttotal: 2.28s\tremaining: 1m 40s\n",
            "221:\tlearn: 0.7587060\ttest: 1.2558899\tbest: 1.2558899 (221)\ttotal: 2.29s\tremaining: 1m 40s\n",
            "222:\tlearn: 0.7570054\ttest: 1.2553312\tbest: 1.2553312 (222)\ttotal: 2.3s\tremaining: 1m 40s\n",
            "223:\tlearn: 0.7551922\ttest: 1.2555298\tbest: 1.2553312 (222)\ttotal: 2.31s\tremaining: 1m 40s\n",
            "224:\tlearn: 0.7540031\ttest: 1.2552685\tbest: 1.2552685 (224)\ttotal: 2.32s\tremaining: 1m 40s\n",
            "225:\tlearn: 0.7525687\ttest: 1.2553013\tbest: 1.2552685 (224)\ttotal: 2.33s\tremaining: 1m 40s\n",
            "226:\tlearn: 0.7505060\ttest: 1.2557853\tbest: 1.2552685 (224)\ttotal: 2.34s\tremaining: 1m 40s\n",
            "227:\tlearn: 0.7492594\ttest: 1.2562963\tbest: 1.2552685 (224)\ttotal: 2.36s\tremaining: 1m 41s\n",
            "228:\tlearn: 0.7475056\ttest: 1.2569911\tbest: 1.2552685 (224)\ttotal: 2.37s\tremaining: 1m 41s\n",
            "229:\tlearn: 0.7455104\ttest: 1.2583533\tbest: 1.2552685 (224)\ttotal: 2.38s\tremaining: 1m 41s\n",
            "230:\tlearn: 0.7442919\ttest: 1.2580017\tbest: 1.2552685 (224)\ttotal: 2.39s\tremaining: 1m 41s\n",
            "231:\tlearn: 0.7422795\ttest: 1.2578905\tbest: 1.2552685 (224)\ttotal: 2.4s\tremaining: 1m 41s\n",
            "232:\tlearn: 0.7408280\ttest: 1.2572217\tbest: 1.2552685 (224)\ttotal: 2.41s\tremaining: 1m 40s\n",
            "233:\tlearn: 0.7393667\ttest: 1.2575490\tbest: 1.2552685 (224)\ttotal: 2.42s\tremaining: 1m 40s\n",
            "234:\tlearn: 0.7381728\ttest: 1.2572250\tbest: 1.2552685 (224)\ttotal: 2.43s\tremaining: 1m 41s\n",
            "235:\tlearn: 0.7366261\ttest: 1.2568001\tbest: 1.2552685 (224)\ttotal: 2.44s\tremaining: 1m 41s\n",
            "236:\tlearn: 0.7352223\ttest: 1.2576137\tbest: 1.2552685 (224)\ttotal: 2.46s\tremaining: 1m 41s\n",
            "237:\tlearn: 0.7339804\ttest: 1.2579111\tbest: 1.2552685 (224)\ttotal: 2.46s\tremaining: 1m 41s\n",
            "238:\tlearn: 0.7330091\ttest: 1.2569070\tbest: 1.2552685 (224)\ttotal: 2.48s\tremaining: 1m 41s\n",
            "239:\tlearn: 0.7312274\ttest: 1.2555331\tbest: 1.2552685 (224)\ttotal: 2.49s\tremaining: 1m 41s\n",
            "240:\tlearn: 0.7296830\ttest: 1.2553133\tbest: 1.2552685 (224)\ttotal: 2.5s\tremaining: 1m 41s\n",
            "241:\tlearn: 0.7280634\ttest: 1.2551601\tbest: 1.2551601 (241)\ttotal: 2.51s\tremaining: 1m 41s\n",
            "242:\tlearn: 0.7267673\ttest: 1.2553246\tbest: 1.2551601 (241)\ttotal: 2.52s\tremaining: 1m 41s\n",
            "243:\tlearn: 0.7251192\ttest: 1.2558261\tbest: 1.2551601 (241)\ttotal: 2.53s\tremaining: 1m 41s\n",
            "244:\tlearn: 0.7237346\ttest: 1.2559524\tbest: 1.2551601 (241)\ttotal: 2.54s\tremaining: 1m 41s\n",
            "245:\tlearn: 0.7222825\ttest: 1.2556607\tbest: 1.2551601 (241)\ttotal: 2.55s\tremaining: 1m 41s\n",
            "246:\tlearn: 0.7214251\ttest: 1.2555406\tbest: 1.2551601 (241)\ttotal: 2.56s\tremaining: 1m 41s\n",
            "247:\tlearn: 0.7198610\ttest: 1.2553832\tbest: 1.2551601 (241)\ttotal: 2.57s\tremaining: 1m 41s\n",
            "248:\tlearn: 0.7178072\ttest: 1.2560250\tbest: 1.2551601 (241)\ttotal: 2.58s\tremaining: 1m 41s\n",
            "249:\tlearn: 0.7167568\ttest: 1.2564602\tbest: 1.2551601 (241)\ttotal: 2.59s\tremaining: 1m 41s\n",
            "250:\tlearn: 0.7150199\ttest: 1.2571501\tbest: 1.2551601 (241)\ttotal: 2.6s\tremaining: 1m 41s\n",
            "251:\tlearn: 0.7133950\ttest: 1.2574426\tbest: 1.2551601 (241)\ttotal: 2.61s\tremaining: 1m 41s\n",
            "252:\tlearn: 0.7120034\ttest: 1.2572583\tbest: 1.2551601 (241)\ttotal: 2.62s\tremaining: 1m 41s\n",
            "253:\tlearn: 0.7097996\ttest: 1.2570495\tbest: 1.2551601 (241)\ttotal: 2.64s\tremaining: 1m 41s\n",
            "254:\tlearn: 0.7083161\ttest: 1.2570307\tbest: 1.2551601 (241)\ttotal: 2.65s\tremaining: 1m 41s\n",
            "255:\tlearn: 0.7068371\ttest: 1.2564654\tbest: 1.2551601 (241)\ttotal: 2.66s\tremaining: 1m 41s\n",
            "256:\tlearn: 0.7056682\ttest: 1.2559045\tbest: 1.2551601 (241)\ttotal: 2.67s\tremaining: 1m 41s\n",
            "257:\tlearn: 0.7040473\ttest: 1.2562835\tbest: 1.2551601 (241)\ttotal: 2.68s\tremaining: 1m 41s\n",
            "258:\tlearn: 0.7024550\ttest: 1.2561208\tbest: 1.2551601 (241)\ttotal: 2.69s\tremaining: 1m 41s\n",
            "259:\tlearn: 0.7014933\ttest: 1.2554828\tbest: 1.2551601 (241)\ttotal: 2.7s\tremaining: 1m 41s\n",
            "260:\tlearn: 0.7004723\ttest: 1.2554398\tbest: 1.2551601 (241)\ttotal: 2.71s\tremaining: 1m 41s\n",
            "261:\tlearn: 0.6993995\ttest: 1.2549971\tbest: 1.2549971 (261)\ttotal: 2.72s\tremaining: 1m 41s\n",
            "262:\tlearn: 0.6976113\ttest: 1.2554752\tbest: 1.2549971 (261)\ttotal: 2.73s\tremaining: 1m 41s\n",
            "263:\tlearn: 0.6964657\ttest: 1.2553834\tbest: 1.2549971 (261)\ttotal: 2.74s\tremaining: 1m 41s\n",
            "264:\tlearn: 0.6951344\ttest: 1.2553031\tbest: 1.2549971 (261)\ttotal: 2.75s\tremaining: 1m 41s\n",
            "265:\tlearn: 0.6935687\ttest: 1.2554489\tbest: 1.2549971 (261)\ttotal: 2.76s\tremaining: 1m 41s\n",
            "266:\tlearn: 0.6919525\ttest: 1.2561766\tbest: 1.2549971 (261)\ttotal: 2.77s\tremaining: 1m 41s\n",
            "267:\tlearn: 0.6903495\ttest: 1.2566263\tbest: 1.2549971 (261)\ttotal: 2.78s\tremaining: 1m 41s\n",
            "268:\tlearn: 0.6893628\ttest: 1.2558721\tbest: 1.2549971 (261)\ttotal: 2.79s\tremaining: 1m 41s\n",
            "269:\tlearn: 0.6877160\ttest: 1.2551320\tbest: 1.2549971 (261)\ttotal: 2.8s\tremaining: 1m 40s\n",
            "270:\tlearn: 0.6861126\ttest: 1.2543103\tbest: 1.2543103 (270)\ttotal: 2.81s\tremaining: 1m 40s\n",
            "271:\tlearn: 0.6849210\ttest: 1.2549408\tbest: 1.2543103 (270)\ttotal: 2.82s\tremaining: 1m 40s\n",
            "272:\tlearn: 0.6837952\ttest: 1.2550955\tbest: 1.2543103 (270)\ttotal: 2.83s\tremaining: 1m 40s\n",
            "273:\tlearn: 0.6825235\ttest: 1.2559280\tbest: 1.2543103 (270)\ttotal: 2.85s\tremaining: 1m 41s\n",
            "274:\tlearn: 0.6808873\ttest: 1.2564047\tbest: 1.2543103 (270)\ttotal: 2.86s\tremaining: 1m 41s\n",
            "275:\tlearn: 0.6795015\ttest: 1.2564887\tbest: 1.2543103 (270)\ttotal: 2.87s\tremaining: 1m 41s\n",
            "276:\tlearn: 0.6779237\ttest: 1.2561164\tbest: 1.2543103 (270)\ttotal: 2.88s\tremaining: 1m 40s\n",
            "277:\tlearn: 0.6768048\ttest: 1.2565943\tbest: 1.2543103 (270)\ttotal: 2.89s\tremaining: 1m 41s\n",
            "278:\tlearn: 0.6759446\ttest: 1.2559742\tbest: 1.2543103 (270)\ttotal: 2.9s\tremaining: 1m 41s\n",
            "279:\tlearn: 0.6748293\ttest: 1.2554678\tbest: 1.2543103 (270)\ttotal: 2.91s\tremaining: 1m 41s\n",
            "280:\tlearn: 0.6739173\ttest: 1.2559610\tbest: 1.2543103 (270)\ttotal: 2.92s\tremaining: 1m 41s\n",
            "281:\tlearn: 0.6724093\ttest: 1.2568693\tbest: 1.2543103 (270)\ttotal: 2.93s\tremaining: 1m 41s\n",
            "282:\tlearn: 0.6712303\ttest: 1.2565651\tbest: 1.2543103 (270)\ttotal: 2.94s\tremaining: 1m 40s\n",
            "283:\tlearn: 0.6694152\ttest: 1.2564923\tbest: 1.2543103 (270)\ttotal: 2.95s\tremaining: 1m 40s\n",
            "284:\tlearn: 0.6683671\ttest: 1.2571567\tbest: 1.2543103 (270)\ttotal: 2.96s\tremaining: 1m 41s\n",
            "285:\tlearn: 0.6676345\ttest: 1.2567770\tbest: 1.2543103 (270)\ttotal: 2.97s\tremaining: 1m 40s\n",
            "286:\tlearn: 0.6666951\ttest: 1.2567721\tbest: 1.2543103 (270)\ttotal: 2.98s\tremaining: 1m 40s\n",
            "287:\tlearn: 0.6654746\ttest: 1.2564335\tbest: 1.2543103 (270)\ttotal: 2.99s\tremaining: 1m 40s\n",
            "288:\tlearn: 0.6641888\ttest: 1.2575681\tbest: 1.2543103 (270)\ttotal: 3s\tremaining: 1m 40s\n",
            "289:\tlearn: 0.6629786\ttest: 1.2576743\tbest: 1.2543103 (270)\ttotal: 3.01s\tremaining: 1m 40s\n",
            "290:\tlearn: 0.6620726\ttest: 1.2581508\tbest: 1.2543103 (270)\ttotal: 3.02s\tremaining: 1m 40s\n",
            "291:\tlearn: 0.6608228\ttest: 1.2585129\tbest: 1.2543103 (270)\ttotal: 3.03s\tremaining: 1m 40s\n",
            "292:\tlearn: 0.6598289\ttest: 1.2580219\tbest: 1.2543103 (270)\ttotal: 3.04s\tremaining: 1m 40s\n",
            "293:\tlearn: 0.6583750\ttest: 1.2581612\tbest: 1.2543103 (270)\ttotal: 3.06s\tremaining: 1m 41s\n",
            "294:\tlearn: 0.6565666\ttest: 1.2581361\tbest: 1.2543103 (270)\ttotal: 3.07s\tremaining: 1m 41s\n",
            "295:\tlearn: 0.6553495\ttest: 1.2587235\tbest: 1.2543103 (270)\ttotal: 3.08s\tremaining: 1m 41s\n",
            "Stopped by overfitting detector  (25 iterations wait)\n",
            "\n",
            "bestTest = 1.254310335\n",
            "bestIteration = 270\n",
            "\n",
            "Shrink model to first 271 iterations.\n",
            "CPU times: user 9.11 s, sys: 633 ms, total: 9.74 s\n",
            "Wall time: 5.54 s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mX6eYy8U4SrB",
        "colab_type": "text"
      },
      "source": [
        "# LightGBM S-Learner"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nfbL1jRd4UP-",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "lgbm_kwargs = dict(\n",
        "    boosting_type='gbdt',\n",
        "    n_estimators=int(10e3),\n",
        "    learning_rate=0.1,\n",
        ")\n",
        "\n",
        "light_model = LGBMRegressor(**lgbm_kwargs)\n",
        "light_dr_model = LGBMRegressor(**lgbm_kwargs)"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8Zz6sAYm9QlY",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "97bae6f6-6255-48e6-cc30-8ce3ccfa251b"
      },
      "source": [
        "%%timeit\n",
        "light_model.fit(\n",
        "    x_train, y_train,\n",
        "    sample_weight=ω(g_train, t_train),\n",
        "    eval_set=(x_test, y_test),\n",
        "    early_stopping_rounds=25,\n",
        "    verbose=False\n",
        ")"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1 loop, best of 3: 3.83 s per loop\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "D53u-0OW9Nnr",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "c1c882b6-b363-4317-d6c1-4779995e1906"
      },
      "source": [
        "%%time\n",
        "light_model.fit(\n",
        "    x_train, y_train,\n",
        "    eval_set=(x_test, y_test),\n",
        "    early_stopping_rounds=25,\n",
        ")\n",
        "\n",
        "light_dr_model.fit(\n",
        "    x_train, y_train,\n",
        "    sample_weight=ω(g_train, t_train),\n",
        "    eval_set=(x_test, y_test),\n",
        "    early_stopping_rounds=25,\n",
        ")"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[1]\tvalid_0's l2: 30.6655\n",
            "Training until validation scores don't improve for 25 rounds.\n",
            "[2]\tvalid_0's l2: 28.3416\n",
            "[3]\tvalid_0's l2: 26.2987\n",
            "[4]\tvalid_0's l2: 24.4696\n",
            "[5]\tvalid_0's l2: 22.9448\n",
            "[6]\tvalid_0's l2: 21.6899\n",
            "[7]\tvalid_0's l2: 20.4788\n",
            "[8]\tvalid_0's l2: 19.4964\n",
            "[9]\tvalid_0's l2: 18.6506\n",
            "[10]\tvalid_0's l2: 17.8094\n",
            "[11]\tvalid_0's l2: 17.1209\n",
            "[12]\tvalid_0's l2: 16.3886\n",
            "[13]\tvalid_0's l2: 15.7883\n",
            "[14]\tvalid_0's l2: 15.2086\n",
            "[15]\tvalid_0's l2: 14.7063\n",
            "[16]\tvalid_0's l2: 14.222\n",
            "[17]\tvalid_0's l2: 13.8232\n",
            "[18]\tvalid_0's l2: 13.4043\n",
            "[19]\tvalid_0's l2: 13.0366\n",
            "[20]\tvalid_0's l2: 12.6033\n",
            "[21]\tvalid_0's l2: 12.2404\n",
            "[22]\tvalid_0's l2: 11.8794\n",
            "[23]\tvalid_0's l2: 11.4736\n",
            "[24]\tvalid_0's l2: 11.1894\n",
            "[25]\tvalid_0's l2: 10.9351\n",
            "[26]\tvalid_0's l2: 10.6592\n",
            "[27]\tvalid_0's l2: 10.3912\n",
            "[28]\tvalid_0's l2: 10.0917\n",
            "[29]\tvalid_0's l2: 9.84054\n",
            "[30]\tvalid_0's l2: 9.57643\n",
            "[31]\tvalid_0's l2: 9.33464\n",
            "[32]\tvalid_0's l2: 9.10132\n",
            "[33]\tvalid_0's l2: 8.87603\n",
            "[34]\tvalid_0's l2: 8.64234\n",
            "[35]\tvalid_0's l2: 8.41263\n",
            "[36]\tvalid_0's l2: 8.1928\n",
            "[37]\tvalid_0's l2: 7.99761\n",
            "[38]\tvalid_0's l2: 7.82251\n",
            "[39]\tvalid_0's l2: 7.63963\n",
            "[40]\tvalid_0's l2: 7.46022\n",
            "[41]\tvalid_0's l2: 7.32732\n",
            "[42]\tvalid_0's l2: 7.15329\n",
            "[43]\tvalid_0's l2: 7.03507\n",
            "[44]\tvalid_0's l2: 6.88825\n",
            "[45]\tvalid_0's l2: 6.76341\n",
            "[46]\tvalid_0's l2: 6.65398\n",
            "[47]\tvalid_0's l2: 6.53861\n",
            "[48]\tvalid_0's l2: 6.40942\n",
            "[49]\tvalid_0's l2: 6.27396\n",
            "[50]\tvalid_0's l2: 6.16024\n",
            "[51]\tvalid_0's l2: 6.04451\n",
            "[52]\tvalid_0's l2: 5.91101\n",
            "[53]\tvalid_0's l2: 5.82086\n",
            "[54]\tvalid_0's l2: 5.71181\n",
            "[55]\tvalid_0's l2: 5.61715\n",
            "[56]\tvalid_0's l2: 5.50405\n",
            "[57]\tvalid_0's l2: 5.43\n",
            "[58]\tvalid_0's l2: 5.34197\n",
            "[59]\tvalid_0's l2: 5.24294\n",
            "[60]\tvalid_0's l2: 5.1595\n",
            "[61]\tvalid_0's l2: 5.07751\n",
            "[62]\tvalid_0's l2: 4.99201\n",
            "[63]\tvalid_0's l2: 4.92333\n",
            "[64]\tvalid_0's l2: 4.83836\n",
            "[65]\tvalid_0's l2: 4.77531\n",
            "[66]\tvalid_0's l2: 4.72096\n",
            "[67]\tvalid_0's l2: 4.6649\n",
            "[68]\tvalid_0's l2: 4.60302\n",
            "[69]\tvalid_0's l2: 4.53247\n",
            "[70]\tvalid_0's l2: 4.4889\n",
            "[71]\tvalid_0's l2: 4.42979\n",
            "[72]\tvalid_0's l2: 4.37007\n",
            "[73]\tvalid_0's l2: 4.31684\n",
            "[74]\tvalid_0's l2: 4.26947\n",
            "[75]\tvalid_0's l2: 4.22856\n",
            "[76]\tvalid_0's l2: 4.17817\n",
            "[77]\tvalid_0's l2: 4.11808\n",
            "[78]\tvalid_0's l2: 4.06895\n",
            "[79]\tvalid_0's l2: 4.03574\n",
            "[80]\tvalid_0's l2: 4.00395\n",
            "[81]\tvalid_0's l2: 3.94587\n",
            "[82]\tvalid_0's l2: 3.8932\n",
            "[83]\tvalid_0's l2: 3.85664\n",
            "[84]\tvalid_0's l2: 3.81527\n",
            "[85]\tvalid_0's l2: 3.78086\n",
            "[86]\tvalid_0's l2: 3.74675\n",
            "[87]\tvalid_0's l2: 3.71178\n",
            "[88]\tvalid_0's l2: 3.67807\n",
            "[89]\tvalid_0's l2: 3.65055\n",
            "[90]\tvalid_0's l2: 3.62227\n",
            "[91]\tvalid_0's l2: 3.59384\n",
            "[92]\tvalid_0's l2: 3.56787\n",
            "[93]\tvalid_0's l2: 3.53943\n",
            "[94]\tvalid_0's l2: 3.50317\n",
            "[95]\tvalid_0's l2: 3.47232\n",
            "[96]\tvalid_0's l2: 3.43527\n",
            "[97]\tvalid_0's l2: 3.40999\n",
            "[98]\tvalid_0's l2: 3.38352\n",
            "[99]\tvalid_0's l2: 3.35326\n",
            "[100]\tvalid_0's l2: 3.32809\n",
            "[101]\tvalid_0's l2: 3.30731\n",
            "[102]\tvalid_0's l2: 3.28303\n",
            "[103]\tvalid_0's l2: 3.2598\n",
            "[104]\tvalid_0's l2: 3.23875\n",
            "[105]\tvalid_0's l2: 3.2196\n",
            "[106]\tvalid_0's l2: 3.19941\n",
            "[107]\tvalid_0's l2: 3.17025\n",
            "[108]\tvalid_0's l2: 3.15168\n",
            "[109]\tvalid_0's l2: 3.1343\n",
            "[110]\tvalid_0's l2: 3.11428\n",
            "[111]\tvalid_0's l2: 3.09716\n",
            "[112]\tvalid_0's l2: 3.08088\n",
            "[113]\tvalid_0's l2: 3.05594\n",
            "[114]\tvalid_0's l2: 3.0378\n",
            "[115]\tvalid_0's l2: 3.02275\n",
            "[116]\tvalid_0's l2: 3.00522\n",
            "[117]\tvalid_0's l2: 2.98844\n",
            "[118]\tvalid_0's l2: 2.9763\n",
            "[119]\tvalid_0's l2: 2.96206\n",
            "[120]\tvalid_0's l2: 2.94398\n",
            "[121]\tvalid_0's l2: 2.92812\n",
            "[122]\tvalid_0's l2: 2.91608\n",
            "[123]\tvalid_0's l2: 2.90495\n",
            "[124]\tvalid_0's l2: 2.89292\n",
            "[125]\tvalid_0's l2: 2.87973\n",
            "[126]\tvalid_0's l2: 2.8688\n",
            "[127]\tvalid_0's l2: 2.8537\n",
            "[128]\tvalid_0's l2: 2.84049\n",
            "[129]\tvalid_0's l2: 2.82932\n",
            "[130]\tvalid_0's l2: 2.81647\n",
            "[131]\tvalid_0's l2: 2.80804\n",
            "[132]\tvalid_0's l2: 2.80124\n",
            "[133]\tvalid_0's l2: 2.78783\n",
            "[134]\tvalid_0's l2: 2.77727\n",
            "[135]\tvalid_0's l2: 2.7706\n",
            "[136]\tvalid_0's l2: 2.76358\n",
            "[137]\tvalid_0's l2: 2.75394\n",
            "[138]\tvalid_0's l2: 2.74749\n",
            "[139]\tvalid_0's l2: 2.74086\n",
            "[140]\tvalid_0's l2: 2.73298\n",
            "[141]\tvalid_0's l2: 2.72257\n",
            "[142]\tvalid_0's l2: 2.71237\n",
            "[143]\tvalid_0's l2: 2.70218\n",
            "[144]\tvalid_0's l2: 2.6942\n",
            "[145]\tvalid_0's l2: 2.68655\n",
            "[146]\tvalid_0's l2: 2.67708\n",
            "[147]\tvalid_0's l2: 2.6691\n",
            "[148]\tvalid_0's l2: 2.66464\n",
            "[149]\tvalid_0's l2: 2.65863\n",
            "[150]\tvalid_0's l2: 2.65178\n",
            "[151]\tvalid_0's l2: 2.6463\n",
            "[152]\tvalid_0's l2: 2.63978\n",
            "[153]\tvalid_0's l2: 2.63615\n",
            "[154]\tvalid_0's l2: 2.63364\n",
            "[155]\tvalid_0's l2: 2.62761\n",
            "[156]\tvalid_0's l2: 2.62434\n",
            "[157]\tvalid_0's l2: 2.62109\n",
            "[158]\tvalid_0's l2: 2.61654\n",
            "[159]\tvalid_0's l2: 2.61467\n",
            "[160]\tvalid_0's l2: 2.60999\n",
            "[161]\tvalid_0's l2: 2.6062\n",
            "[162]\tvalid_0's l2: 2.60199\n",
            "[163]\tvalid_0's l2: 2.59894\n",
            "[164]\tvalid_0's l2: 2.596\n",
            "[165]\tvalid_0's l2: 2.59492\n",
            "[166]\tvalid_0's l2: 2.59522\n",
            "[167]\tvalid_0's l2: 2.59082\n",
            "[168]\tvalid_0's l2: 2.58863\n",
            "[169]\tvalid_0's l2: 2.58477\n",
            "[170]\tvalid_0's l2: 2.58317\n",
            "[171]\tvalid_0's l2: 2.58191\n",
            "[172]\tvalid_0's l2: 2.58152\n",
            "[173]\tvalid_0's l2: 2.58105\n",
            "[174]\tvalid_0's l2: 2.57784\n",
            "[175]\tvalid_0's l2: 2.57638\n",
            "[176]\tvalid_0's l2: 2.57284\n",
            "[177]\tvalid_0's l2: 2.57271\n",
            "[178]\tvalid_0's l2: 2.56635\n",
            "[179]\tvalid_0's l2: 2.56362\n",
            "[180]\tvalid_0's l2: 2.56346\n",
            "[181]\tvalid_0's l2: 2.56309\n",
            "[182]\tvalid_0's l2: 2.55838\n",
            "[183]\tvalid_0's l2: 2.55597\n",
            "[184]\tvalid_0's l2: 2.55413\n",
            "[185]\tvalid_0's l2: 2.55277\n",
            "[186]\tvalid_0's l2: 2.55073\n",
            "[187]\tvalid_0's l2: 2.54994\n",
            "[188]\tvalid_0's l2: 2.54612\n",
            "[189]\tvalid_0's l2: 2.54307\n",
            "[190]\tvalid_0's l2: 2.54058\n",
            "[191]\tvalid_0's l2: 2.53548\n",
            "[192]\tvalid_0's l2: 2.53305\n",
            "[193]\tvalid_0's l2: 2.53148\n",
            "[194]\tvalid_0's l2: 2.5304\n",
            "[195]\tvalid_0's l2: 2.52918\n",
            "[196]\tvalid_0's l2: 2.52826\n",
            "[197]\tvalid_0's l2: 2.52703\n",
            "[198]\tvalid_0's l2: 2.52459\n",
            "[199]\tvalid_0's l2: 2.52419\n",
            "[200]\tvalid_0's l2: 2.5244\n",
            "[201]\tvalid_0's l2: 2.52276\n",
            "[202]\tvalid_0's l2: 2.52005\n",
            "[203]\tvalid_0's l2: 2.5204\n",
            "[204]\tvalid_0's l2: 2.51998\n",
            "[205]\tvalid_0's l2: 2.51826\n",
            "[206]\tvalid_0's l2: 2.51922\n",
            "[207]\tvalid_0's l2: 2.51813\n",
            "[208]\tvalid_0's l2: 2.51647\n",
            "[209]\tvalid_0's l2: 2.51536\n",
            "[210]\tvalid_0's l2: 2.5152\n",
            "[211]\tvalid_0's l2: 2.51417\n",
            "[212]\tvalid_0's l2: 2.51195\n",
            "[213]\tvalid_0's l2: 2.51005\n",
            "[214]\tvalid_0's l2: 2.51011\n",
            "[215]\tvalid_0's l2: 2.50847\n",
            "[216]\tvalid_0's l2: 2.50785\n",
            "[217]\tvalid_0's l2: 2.50556\n",
            "[218]\tvalid_0's l2: 2.50639\n",
            "[219]\tvalid_0's l2: 2.50708\n",
            "[220]\tvalid_0's l2: 2.50613\n",
            "[221]\tvalid_0's l2: 2.50487\n",
            "[222]\tvalid_0's l2: 2.50306\n",
            "[223]\tvalid_0's l2: 2.50063\n",
            "[224]\tvalid_0's l2: 2.49948\n",
            "[225]\tvalid_0's l2: 2.49908\n",
            "[226]\tvalid_0's l2: 2.49971\n",
            "[227]\tvalid_0's l2: 2.49763\n",
            "[228]\tvalid_0's l2: 2.49666\n",
            "[229]\tvalid_0's l2: 2.49817\n",
            "[230]\tvalid_0's l2: 2.49804\n",
            "[231]\tvalid_0's l2: 2.49918\n",
            "[232]\tvalid_0's l2: 2.50092\n",
            "[233]\tvalid_0's l2: 2.50205\n",
            "[234]\tvalid_0's l2: 2.50178\n",
            "[235]\tvalid_0's l2: 2.501\n",
            "[236]\tvalid_0's l2: 2.50064\n",
            "[237]\tvalid_0's l2: 2.49819\n",
            "[238]\tvalid_0's l2: 2.49769\n",
            "[239]\tvalid_0's l2: 2.49738\n",
            "[240]\tvalid_0's l2: 2.49568\n",
            "[241]\tvalid_0's l2: 2.49531\n",
            "[242]\tvalid_0's l2: 2.49649\n",
            "[243]\tvalid_0's l2: 2.49702\n",
            "[244]\tvalid_0's l2: 2.49461\n",
            "[245]\tvalid_0's l2: 2.49566\n",
            "[246]\tvalid_0's l2: 2.49625\n",
            "[247]\tvalid_0's l2: 2.49408\n",
            "[248]\tvalid_0's l2: 2.49276\n",
            "[249]\tvalid_0's l2: 2.49169\n",
            "[250]\tvalid_0's l2: 2.49133\n",
            "[251]\tvalid_0's l2: 2.49085\n",
            "[252]\tvalid_0's l2: 2.49023\n",
            "[253]\tvalid_0's l2: 2.48855\n",
            "[254]\tvalid_0's l2: 2.48732\n",
            "[255]\tvalid_0's l2: 2.48826\n",
            "[256]\tvalid_0's l2: 2.48859\n",
            "[257]\tvalid_0's l2: 2.48881\n",
            "[258]\tvalid_0's l2: 2.48639\n",
            "[259]\tvalid_0's l2: 2.48459\n",
            "[260]\tvalid_0's l2: 2.4848\n",
            "[261]\tvalid_0's l2: 2.48373\n",
            "[262]\tvalid_0's l2: 2.48149\n",
            "[263]\tvalid_0's l2: 2.48026\n",
            "[264]\tvalid_0's l2: 2.4804\n",
            "[265]\tvalid_0's l2: 2.47978\n",
            "[266]\tvalid_0's l2: 2.48051\n",
            "[267]\tvalid_0's l2: 2.48094\n",
            "[268]\tvalid_0's l2: 2.48153\n",
            "[269]\tvalid_0's l2: 2.48029\n",
            "[270]\tvalid_0's l2: 2.47843\n",
            "[271]\tvalid_0's l2: 2.47791\n",
            "[272]\tvalid_0's l2: 2.47715\n",
            "[273]\tvalid_0's l2: 2.47538\n",
            "[274]\tvalid_0's l2: 2.47388\n",
            "[275]\tvalid_0's l2: 2.47281\n",
            "[276]\tvalid_0's l2: 2.47353\n",
            "[277]\tvalid_0's l2: 2.47373\n",
            "[278]\tvalid_0's l2: 2.47342\n",
            "[279]\tvalid_0's l2: 2.47243\n",
            "[280]\tvalid_0's l2: 2.47239\n",
            "[281]\tvalid_0's l2: 2.46939\n",
            "[282]\tvalid_0's l2: 2.46735\n",
            "[283]\tvalid_0's l2: 2.46721\n",
            "[284]\tvalid_0's l2: 2.46642\n",
            "[285]\tvalid_0's l2: 2.46574\n",
            "[286]\tvalid_0's l2: 2.46452\n",
            "[287]\tvalid_0's l2: 2.46333\n",
            "[288]\tvalid_0's l2: 2.4631\n",
            "[289]\tvalid_0's l2: 2.46139\n",
            "[290]\tvalid_0's l2: 2.46218\n",
            "[291]\tvalid_0's l2: 2.45934\n",
            "[292]\tvalid_0's l2: 2.45945\n",
            "[293]\tvalid_0's l2: 2.45866\n",
            "[294]\tvalid_0's l2: 2.4584\n",
            "[295]\tvalid_0's l2: 2.45834\n",
            "[296]\tvalid_0's l2: 2.45765\n",
            "[297]\tvalid_0's l2: 2.45822\n",
            "[298]\tvalid_0's l2: 2.45947\n",
            "[299]\tvalid_0's l2: 2.45974\n",
            "[300]\tvalid_0's l2: 2.45972\n",
            "[301]\tvalid_0's l2: 2.46002\n",
            "[302]\tvalid_0's l2: 2.45999\n",
            "[303]\tvalid_0's l2: 2.45997\n",
            "[304]\tvalid_0's l2: 2.46041\n",
            "[305]\tvalid_0's l2: 2.4596\n",
            "[306]\tvalid_0's l2: 2.45956\n",
            "[307]\tvalid_0's l2: 2.46034\n",
            "[308]\tvalid_0's l2: 2.46069\n",
            "[309]\tvalid_0's l2: 2.45985\n",
            "[310]\tvalid_0's l2: 2.45932\n",
            "[311]\tvalid_0's l2: 2.45883\n",
            "[312]\tvalid_0's l2: 2.45737\n",
            "[313]\tvalid_0's l2: 2.45697\n",
            "[314]\tvalid_0's l2: 2.45678\n",
            "[315]\tvalid_0's l2: 2.45833\n",
            "[316]\tvalid_0's l2: 2.45781\n",
            "[317]\tvalid_0's l2: 2.45768\n",
            "[318]\tvalid_0's l2: 2.45711\n",
            "[319]\tvalid_0's l2: 2.45781\n",
            "[320]\tvalid_0's l2: 2.4571\n",
            "[321]\tvalid_0's l2: 2.45527\n",
            "[322]\tvalid_0's l2: 2.45484\n",
            "[323]\tvalid_0's l2: 2.45595\n",
            "[324]\tvalid_0's l2: 2.45518\n",
            "[325]\tvalid_0's l2: 2.45496\n",
            "[326]\tvalid_0's l2: 2.45391\n",
            "[327]\tvalid_0's l2: 2.45342\n",
            "[328]\tvalid_0's l2: 2.45334\n",
            "[329]\tvalid_0's l2: 2.4541\n",
            "[330]\tvalid_0's l2: 2.4544\n",
            "[331]\tvalid_0's l2: 2.45331\n",
            "[332]\tvalid_0's l2: 2.45325\n",
            "[333]\tvalid_0's l2: 2.45311\n",
            "[334]\tvalid_0's l2: 2.45356\n",
            "[335]\tvalid_0's l2: 2.45209\n",
            "[336]\tvalid_0's l2: 2.45222\n",
            "[337]\tvalid_0's l2: 2.45257\n",
            "[338]\tvalid_0's l2: 2.45356\n",
            "[339]\tvalid_0's l2: 2.45287\n",
            "[340]\tvalid_0's l2: 2.45304\n",
            "[341]\tvalid_0's l2: 2.45092\n",
            "[342]\tvalid_0's l2: 2.45008\n",
            "[343]\tvalid_0's l2: 2.45048\n",
            "[344]\tvalid_0's l2: 2.44945\n",
            "[345]\tvalid_0's l2: 2.44933\n",
            "[346]\tvalid_0's l2: 2.4495\n",
            "[347]\tvalid_0's l2: 2.44917\n",
            "[348]\tvalid_0's l2: 2.4499\n",
            "[349]\tvalid_0's l2: 2.44916\n",
            "[350]\tvalid_0's l2: 2.44793\n",
            "[351]\tvalid_0's l2: 2.449\n",
            "[352]\tvalid_0's l2: 2.44916\n",
            "[353]\tvalid_0's l2: 2.44747\n",
            "[354]\tvalid_0's l2: 2.44778\n",
            "[355]\tvalid_0's l2: 2.44725\n",
            "[356]\tvalid_0's l2: 2.4466\n",
            "[357]\tvalid_0's l2: 2.44516\n",
            "[358]\tvalid_0's l2: 2.44391\n",
            "[359]\tvalid_0's l2: 2.44561\n",
            "[360]\tvalid_0's l2: 2.44529\n",
            "[361]\tvalid_0's l2: 2.44542\n",
            "[362]\tvalid_0's l2: 2.44367\n",
            "[363]\tvalid_0's l2: 2.44418\n",
            "[364]\tvalid_0's l2: 2.44364\n",
            "[365]\tvalid_0's l2: 2.44345\n",
            "[366]\tvalid_0's l2: 2.44269\n",
            "[367]\tvalid_0's l2: 2.4432\n",
            "[368]\tvalid_0's l2: 2.44201\n",
            "[369]\tvalid_0's l2: 2.44089\n",
            "[370]\tvalid_0's l2: 2.44148\n",
            "[371]\tvalid_0's l2: 2.44223\n",
            "[372]\tvalid_0's l2: 2.4412\n",
            "[373]\tvalid_0's l2: 2.44027\n",
            "[374]\tvalid_0's l2: 2.4397\n",
            "[375]\tvalid_0's l2: 2.43983\n",
            "[376]\tvalid_0's l2: 2.43936\n",
            "[377]\tvalid_0's l2: 2.44009\n",
            "[378]\tvalid_0's l2: 2.44079\n",
            "[379]\tvalid_0's l2: 2.43977\n",
            "[380]\tvalid_0's l2: 2.44024\n",
            "[381]\tvalid_0's l2: 2.43999\n",
            "[382]\tvalid_0's l2: 2.43934\n",
            "[383]\tvalid_0's l2: 2.43884\n",
            "[384]\tvalid_0's l2: 2.43874\n",
            "[385]\tvalid_0's l2: 2.4376\n",
            "[386]\tvalid_0's l2: 2.43687\n",
            "[387]\tvalid_0's l2: 2.43827\n",
            "[388]\tvalid_0's l2: 2.43925\n",
            "[389]\tvalid_0's l2: 2.44011\n",
            "[390]\tvalid_0's l2: 2.43971\n",
            "[391]\tvalid_0's l2: 2.43866\n",
            "[392]\tvalid_0's l2: 2.4392\n",
            "[393]\tvalid_0's l2: 2.43854\n",
            "[394]\tvalid_0's l2: 2.43914\n",
            "[395]\tvalid_0's l2: 2.43836\n",
            "[396]\tvalid_0's l2: 2.43825\n",
            "[397]\tvalid_0's l2: 2.43805\n",
            "[398]\tvalid_0's l2: 2.43782\n",
            "[399]\tvalid_0's l2: 2.43845\n",
            "[400]\tvalid_0's l2: 2.43858\n",
            "[401]\tvalid_0's l2: 2.43781\n",
            "[402]\tvalid_0's l2: 2.43841\n",
            "[403]\tvalid_0's l2: 2.43868\n",
            "[404]\tvalid_0's l2: 2.43901\n",
            "[405]\tvalid_0's l2: 2.43868\n",
            "[406]\tvalid_0's l2: 2.43887\n",
            "[407]\tvalid_0's l2: 2.43947\n",
            "[408]\tvalid_0's l2: 2.43926\n",
            "[409]\tvalid_0's l2: 2.44041\n",
            "[410]\tvalid_0's l2: 2.43999\n",
            "[411]\tvalid_0's l2: 2.4399\n",
            "Early stopping, best iteration is:\n",
            "[386]\tvalid_0's l2: 2.43687\n",
            "[1]\tvalid_0's l2: 30.7049\n",
            "Training until validation scores don't improve for 25 rounds.\n",
            "[2]\tvalid_0's l2: 28.2098\n",
            "[3]\tvalid_0's l2: 26.269\n",
            "[4]\tvalid_0's l2: 24.4773\n",
            "[5]\tvalid_0's l2: 22.9852\n",
            "[6]\tvalid_0's l2: 21.6737\n",
            "[7]\tvalid_0's l2: 20.5543\n",
            "[8]\tvalid_0's l2: 19.679\n",
            "[9]\tvalid_0's l2: 18.8293\n",
            "[10]\tvalid_0's l2: 17.9434\n",
            "[11]\tvalid_0's l2: 17.1579\n",
            "[12]\tvalid_0's l2: 16.5013\n",
            "[13]\tvalid_0's l2: 15.8822\n",
            "[14]\tvalid_0's l2: 15.3128\n",
            "[15]\tvalid_0's l2: 14.7733\n",
            "[16]\tvalid_0's l2: 14.2901\n",
            "[17]\tvalid_0's l2: 13.8144\n",
            "[18]\tvalid_0's l2: 13.4577\n",
            "[19]\tvalid_0's l2: 12.9975\n",
            "[20]\tvalid_0's l2: 12.532\n",
            "[21]\tvalid_0's l2: 12.1782\n",
            "[22]\tvalid_0's l2: 11.8082\n",
            "[23]\tvalid_0's l2: 11.4854\n",
            "[24]\tvalid_0's l2: 11.1894\n",
            "[25]\tvalid_0's l2: 10.8867\n",
            "[26]\tvalid_0's l2: 10.568\n",
            "[27]\tvalid_0's l2: 10.3374\n",
            "[28]\tvalid_0's l2: 10.1031\n",
            "[29]\tvalid_0's l2: 9.86306\n",
            "[30]\tvalid_0's l2: 9.62281\n",
            "[31]\tvalid_0's l2: 9.40389\n",
            "[32]\tvalid_0's l2: 9.17389\n",
            "[33]\tvalid_0's l2: 8.93741\n",
            "[34]\tvalid_0's l2: 8.70525\n",
            "[35]\tvalid_0's l2: 8.48118\n",
            "[36]\tvalid_0's l2: 8.31354\n",
            "[37]\tvalid_0's l2: 8.1431\n",
            "[38]\tvalid_0's l2: 7.95554\n",
            "[39]\tvalid_0's l2: 7.7617\n",
            "[40]\tvalid_0's l2: 7.61152\n",
            "[41]\tvalid_0's l2: 7.44942\n",
            "[42]\tvalid_0's l2: 7.28906\n",
            "[43]\tvalid_0's l2: 7.11932\n",
            "[44]\tvalid_0's l2: 6.99261\n",
            "[45]\tvalid_0's l2: 6.86349\n",
            "[46]\tvalid_0's l2: 6.70643\n",
            "[47]\tvalid_0's l2: 6.59039\n",
            "[48]\tvalid_0's l2: 6.45828\n",
            "[49]\tvalid_0's l2: 6.36499\n",
            "[50]\tvalid_0's l2: 6.25954\n",
            "[51]\tvalid_0's l2: 6.12563\n",
            "[52]\tvalid_0's l2: 6.00812\n",
            "[53]\tvalid_0's l2: 5.89106\n",
            "[54]\tvalid_0's l2: 5.79011\n",
            "[55]\tvalid_0's l2: 5.70882\n",
            "[56]\tvalid_0's l2: 5.61529\n",
            "[57]\tvalid_0's l2: 5.5248\n",
            "[58]\tvalid_0's l2: 5.44191\n",
            "[59]\tvalid_0's l2: 5.35638\n",
            "[60]\tvalid_0's l2: 5.28488\n",
            "[61]\tvalid_0's l2: 5.21763\n",
            "[62]\tvalid_0's l2: 5.12197\n",
            "[63]\tvalid_0's l2: 5.03325\n",
            "[64]\tvalid_0's l2: 4.9621\n",
            "[65]\tvalid_0's l2: 4.9\n",
            "[66]\tvalid_0's l2: 4.82725\n",
            "[67]\tvalid_0's l2: 4.76036\n",
            "[68]\tvalid_0's l2: 4.69426\n",
            "[69]\tvalid_0's l2: 4.62883\n",
            "[70]\tvalid_0's l2: 4.56332\n",
            "[71]\tvalid_0's l2: 4.50036\n",
            "[72]\tvalid_0's l2: 4.45546\n",
            "[73]\tvalid_0's l2: 4.39238\n",
            "[74]\tvalid_0's l2: 4.33028\n",
            "[75]\tvalid_0's l2: 4.28686\n",
            "[76]\tvalid_0's l2: 4.23147\n",
            "[77]\tvalid_0's l2: 4.18534\n",
            "[78]\tvalid_0's l2: 4.13416\n",
            "[79]\tvalid_0's l2: 4.08964\n",
            "[80]\tvalid_0's l2: 4.04523\n",
            "[81]\tvalid_0's l2: 4.00407\n",
            "[82]\tvalid_0's l2: 3.96677\n",
            "[83]\tvalid_0's l2: 3.9243\n",
            "[84]\tvalid_0's l2: 3.8871\n",
            "[85]\tvalid_0's l2: 3.84188\n",
            "[86]\tvalid_0's l2: 3.80188\n",
            "[87]\tvalid_0's l2: 3.76237\n",
            "[88]\tvalid_0's l2: 3.73179\n",
            "[89]\tvalid_0's l2: 3.69456\n",
            "[90]\tvalid_0's l2: 3.66076\n",
            "[91]\tvalid_0's l2: 3.62658\n",
            "[92]\tvalid_0's l2: 3.59901\n",
            "[93]\tvalid_0's l2: 3.56315\n",
            "[94]\tvalid_0's l2: 3.52998\n",
            "[95]\tvalid_0's l2: 3.49372\n",
            "[96]\tvalid_0's l2: 3.46682\n",
            "[97]\tvalid_0's l2: 3.4444\n",
            "[98]\tvalid_0's l2: 3.41347\n",
            "[99]\tvalid_0's l2: 3.3881\n",
            "[100]\tvalid_0's l2: 3.36089\n",
            "[101]\tvalid_0's l2: 3.33579\n",
            "[102]\tvalid_0's l2: 3.31395\n",
            "[103]\tvalid_0's l2: 3.2869\n",
            "[104]\tvalid_0's l2: 3.26926\n",
            "[105]\tvalid_0's l2: 3.23956\n",
            "[106]\tvalid_0's l2: 3.21949\n",
            "[107]\tvalid_0's l2: 3.19607\n",
            "[108]\tvalid_0's l2: 3.17713\n",
            "[109]\tvalid_0's l2: 3.16273\n",
            "[110]\tvalid_0's l2: 3.14316\n",
            "[111]\tvalid_0's l2: 3.11896\n",
            "[112]\tvalid_0's l2: 3.10365\n",
            "[113]\tvalid_0's l2: 3.08694\n",
            "[114]\tvalid_0's l2: 3.06355\n",
            "[115]\tvalid_0's l2: 3.04657\n",
            "[116]\tvalid_0's l2: 3.02863\n",
            "[117]\tvalid_0's l2: 3.0088\n",
            "[118]\tvalid_0's l2: 2.99193\n",
            "[119]\tvalid_0's l2: 2.98121\n",
            "[120]\tvalid_0's l2: 2.97013\n",
            "[121]\tvalid_0's l2: 2.95311\n",
            "[122]\tvalid_0's l2: 2.9367\n",
            "[123]\tvalid_0's l2: 2.91728\n",
            "[124]\tvalid_0's l2: 2.9099\n",
            "[125]\tvalid_0's l2: 2.89094\n",
            "[126]\tvalid_0's l2: 2.87748\n",
            "[127]\tvalid_0's l2: 2.86293\n",
            "[128]\tvalid_0's l2: 2.85038\n",
            "[129]\tvalid_0's l2: 2.83902\n",
            "[130]\tvalid_0's l2: 2.82656\n",
            "[131]\tvalid_0's l2: 2.81614\n",
            "[132]\tvalid_0's l2: 2.80317\n",
            "[133]\tvalid_0's l2: 2.78999\n",
            "[134]\tvalid_0's l2: 2.77696\n",
            "[135]\tvalid_0's l2: 2.76665\n",
            "[136]\tvalid_0's l2: 2.75351\n",
            "[137]\tvalid_0's l2: 2.739\n",
            "[138]\tvalid_0's l2: 2.72772\n",
            "[139]\tvalid_0's l2: 2.7221\n",
            "[140]\tvalid_0's l2: 2.71325\n",
            "[141]\tvalid_0's l2: 2.70277\n",
            "[142]\tvalid_0's l2: 2.69311\n",
            "[143]\tvalid_0's l2: 2.68142\n",
            "[144]\tvalid_0's l2: 2.6747\n",
            "[145]\tvalid_0's l2: 2.66918\n",
            "[146]\tvalid_0's l2: 2.66343\n",
            "[147]\tvalid_0's l2: 2.65693\n",
            "[148]\tvalid_0's l2: 2.64653\n",
            "[149]\tvalid_0's l2: 2.64103\n",
            "[150]\tvalid_0's l2: 2.63433\n",
            "[151]\tvalid_0's l2: 2.62892\n",
            "[152]\tvalid_0's l2: 2.62205\n",
            "[153]\tvalid_0's l2: 2.61569\n",
            "[154]\tvalid_0's l2: 2.61212\n",
            "[155]\tvalid_0's l2: 2.60599\n",
            "[156]\tvalid_0's l2: 2.60141\n",
            "[157]\tvalid_0's l2: 2.59666\n",
            "[158]\tvalid_0's l2: 2.59073\n",
            "[159]\tvalid_0's l2: 2.586\n",
            "[160]\tvalid_0's l2: 2.58186\n",
            "[161]\tvalid_0's l2: 2.57694\n",
            "[162]\tvalid_0's l2: 2.57111\n",
            "[163]\tvalid_0's l2: 2.56682\n",
            "[164]\tvalid_0's l2: 2.56421\n",
            "[165]\tvalid_0's l2: 2.56043\n",
            "[166]\tvalid_0's l2: 2.55626\n",
            "[167]\tvalid_0's l2: 2.55114\n",
            "[168]\tvalid_0's l2: 2.54632\n",
            "[169]\tvalid_0's l2: 2.54345\n",
            "[170]\tvalid_0's l2: 2.54058\n",
            "[171]\tvalid_0's l2: 2.53739\n",
            "[172]\tvalid_0's l2: 2.53338\n",
            "[173]\tvalid_0's l2: 2.53051\n",
            "[174]\tvalid_0's l2: 2.52787\n",
            "[175]\tvalid_0's l2: 2.52568\n",
            "[176]\tvalid_0's l2: 2.52346\n",
            "[177]\tvalid_0's l2: 2.52031\n",
            "[178]\tvalid_0's l2: 2.51886\n",
            "[179]\tvalid_0's l2: 2.51554\n",
            "[180]\tvalid_0's l2: 2.51215\n",
            "[181]\tvalid_0's l2: 2.51055\n",
            "[182]\tvalid_0's l2: 2.50858\n",
            "[183]\tvalid_0's l2: 2.50503\n",
            "[184]\tvalid_0's l2: 2.50281\n",
            "[185]\tvalid_0's l2: 2.49868\n",
            "[186]\tvalid_0's l2: 2.49614\n",
            "[187]\tvalid_0's l2: 2.49333\n",
            "[188]\tvalid_0's l2: 2.49287\n",
            "[189]\tvalid_0's l2: 2.4901\n",
            "[190]\tvalid_0's l2: 2.48914\n",
            "[191]\tvalid_0's l2: 2.48677\n",
            "[192]\tvalid_0's l2: 2.48549\n",
            "[193]\tvalid_0's l2: 2.48537\n",
            "[194]\tvalid_0's l2: 2.48424\n",
            "[195]\tvalid_0's l2: 2.48078\n",
            "[196]\tvalid_0's l2: 2.47534\n",
            "[197]\tvalid_0's l2: 2.47488\n",
            "[198]\tvalid_0's l2: 2.47445\n",
            "[199]\tvalid_0's l2: 2.47179\n",
            "[200]\tvalid_0's l2: 2.46949\n",
            "[201]\tvalid_0's l2: 2.46798\n",
            "[202]\tvalid_0's l2: 2.46433\n",
            "[203]\tvalid_0's l2: 2.46291\n",
            "[204]\tvalid_0's l2: 2.46005\n",
            "[205]\tvalid_0's l2: 2.45982\n",
            "[206]\tvalid_0's l2: 2.46059\n",
            "[207]\tvalid_0's l2: 2.4602\n",
            "[208]\tvalid_0's l2: 2.46085\n",
            "[209]\tvalid_0's l2: 2.46001\n",
            "[210]\tvalid_0's l2: 2.45974\n",
            "[211]\tvalid_0's l2: 2.458\n",
            "[212]\tvalid_0's l2: 2.45448\n",
            "[213]\tvalid_0's l2: 2.45323\n",
            "[214]\tvalid_0's l2: 2.45289\n",
            "[215]\tvalid_0's l2: 2.45085\n",
            "[216]\tvalid_0's l2: 2.45074\n",
            "[217]\tvalid_0's l2: 2.44959\n",
            "[218]\tvalid_0's l2: 2.44848\n",
            "[219]\tvalid_0's l2: 2.44788\n",
            "[220]\tvalid_0's l2: 2.44558\n",
            "[221]\tvalid_0's l2: 2.44256\n",
            "[222]\tvalid_0's l2: 2.44229\n",
            "[223]\tvalid_0's l2: 2.44137\n",
            "[224]\tvalid_0's l2: 2.43988\n",
            "[225]\tvalid_0's l2: 2.43846\n",
            "[226]\tvalid_0's l2: 2.43629\n",
            "[227]\tvalid_0's l2: 2.43562\n",
            "[228]\tvalid_0's l2: 2.43209\n",
            "[229]\tvalid_0's l2: 2.43116\n",
            "[230]\tvalid_0's l2: 2.42815\n",
            "[231]\tvalid_0's l2: 2.42943\n",
            "[232]\tvalid_0's l2: 2.42655\n",
            "[233]\tvalid_0's l2: 2.42523\n",
            "[234]\tvalid_0's l2: 2.42174\n",
            "[235]\tvalid_0's l2: 2.42051\n",
            "[236]\tvalid_0's l2: 2.41916\n",
            "[237]\tvalid_0's l2: 2.41817\n",
            "[238]\tvalid_0's l2: 2.41726\n",
            "[239]\tvalid_0's l2: 2.41769\n",
            "[240]\tvalid_0's l2: 2.41695\n",
            "[241]\tvalid_0's l2: 2.41635\n",
            "[242]\tvalid_0's l2: 2.41428\n",
            "[243]\tvalid_0's l2: 2.41458\n",
            "[244]\tvalid_0's l2: 2.41363\n",
            "[245]\tvalid_0's l2: 2.41091\n",
            "[246]\tvalid_0's l2: 2.41006\n",
            "[247]\tvalid_0's l2: 2.40958\n",
            "[248]\tvalid_0's l2: 2.40962\n",
            "[249]\tvalid_0's l2: 2.40795\n",
            "[250]\tvalid_0's l2: 2.40571\n",
            "[251]\tvalid_0's l2: 2.40472\n",
            "[252]\tvalid_0's l2: 2.4043\n",
            "[253]\tvalid_0's l2: 2.40348\n",
            "[254]\tvalid_0's l2: 2.40369\n",
            "[255]\tvalid_0's l2: 2.4026\n",
            "[256]\tvalid_0's l2: 2.40166\n",
            "[257]\tvalid_0's l2: 2.39923\n",
            "[258]\tvalid_0's l2: 2.39962\n",
            "[259]\tvalid_0's l2: 2.39912\n",
            "[260]\tvalid_0's l2: 2.39869\n",
            "[261]\tvalid_0's l2: 2.39668\n",
            "[262]\tvalid_0's l2: 2.39578\n",
            "[263]\tvalid_0's l2: 2.39466\n",
            "[264]\tvalid_0's l2: 2.39437\n",
            "[265]\tvalid_0's l2: 2.39352\n",
            "[266]\tvalid_0's l2: 2.3922\n",
            "[267]\tvalid_0's l2: 2.39223\n",
            "[268]\tvalid_0's l2: 2.39055\n",
            "[269]\tvalid_0's l2: 2.3891\n",
            "[270]\tvalid_0's l2: 2.38711\n",
            "[271]\tvalid_0's l2: 2.38744\n",
            "[272]\tvalid_0's l2: 2.3859\n",
            "[273]\tvalid_0's l2: 2.38416\n",
            "[274]\tvalid_0's l2: 2.38414\n",
            "[275]\tvalid_0's l2: 2.38214\n",
            "[276]\tvalid_0's l2: 2.3819\n",
            "[277]\tvalid_0's l2: 2.38136\n",
            "[278]\tvalid_0's l2: 2.37936\n",
            "[279]\tvalid_0's l2: 2.37946\n",
            "[280]\tvalid_0's l2: 2.37907\n",
            "[281]\tvalid_0's l2: 2.37862\n",
            "[282]\tvalid_0's l2: 2.37857\n",
            "[283]\tvalid_0's l2: 2.37794\n",
            "[284]\tvalid_0's l2: 2.3778\n",
            "[285]\tvalid_0's l2: 2.3766\n",
            "[286]\tvalid_0's l2: 2.3754\n",
            "[287]\tvalid_0's l2: 2.37336\n",
            "[288]\tvalid_0's l2: 2.3726\n",
            "[289]\tvalid_0's l2: 2.37272\n",
            "[290]\tvalid_0's l2: 2.37228\n",
            "[291]\tvalid_0's l2: 2.37\n",
            "[292]\tvalid_0's l2: 2.3699\n",
            "[293]\tvalid_0's l2: 2.36989\n",
            "[294]\tvalid_0's l2: 2.36994\n",
            "[295]\tvalid_0's l2: 2.36889\n",
            "[296]\tvalid_0's l2: 2.37013\n",
            "[297]\tvalid_0's l2: 2.37125\n",
            "[298]\tvalid_0's l2: 2.37018\n",
            "[299]\tvalid_0's l2: 2.37102\n",
            "[300]\tvalid_0's l2: 2.37101\n",
            "[301]\tvalid_0's l2: 2.37027\n",
            "[302]\tvalid_0's l2: 2.37049\n",
            "[303]\tvalid_0's l2: 2.36814\n",
            "[304]\tvalid_0's l2: 2.36697\n",
            "[305]\tvalid_0's l2: 2.36527\n",
            "[306]\tvalid_0's l2: 2.36417\n",
            "[307]\tvalid_0's l2: 2.36339\n",
            "[308]\tvalid_0's l2: 2.3641\n",
            "[309]\tvalid_0's l2: 2.3644\n",
            "[310]\tvalid_0's l2: 2.3629\n",
            "[311]\tvalid_0's l2: 2.36045\n",
            "[312]\tvalid_0's l2: 2.35785\n",
            "[313]\tvalid_0's l2: 2.35699\n",
            "[314]\tvalid_0's l2: 2.35513\n",
            "[315]\tvalid_0's l2: 2.35427\n",
            "[316]\tvalid_0's l2: 2.35379\n",
            "[317]\tvalid_0's l2: 2.35231\n",
            "[318]\tvalid_0's l2: 2.35146\n",
            "[319]\tvalid_0's l2: 2.3518\n",
            "[320]\tvalid_0's l2: 2.35064\n",
            "[321]\tvalid_0's l2: 2.34986\n",
            "[322]\tvalid_0's l2: 2.34871\n",
            "[323]\tvalid_0's l2: 2.34725\n",
            "[324]\tvalid_0's l2: 2.34578\n",
            "[325]\tvalid_0's l2: 2.34443\n",
            "[326]\tvalid_0's l2: 2.34457\n",
            "[327]\tvalid_0's l2: 2.3427\n",
            "[328]\tvalid_0's l2: 2.34249\n",
            "[329]\tvalid_0's l2: 2.34146\n",
            "[330]\tvalid_0's l2: 2.34033\n",
            "[331]\tvalid_0's l2: 2.33872\n",
            "[332]\tvalid_0's l2: 2.33839\n",
            "[333]\tvalid_0's l2: 2.33721\n",
            "[334]\tvalid_0's l2: 2.33803\n",
            "[335]\tvalid_0's l2: 2.33744\n",
            "[336]\tvalid_0's l2: 2.33742\n",
            "[337]\tvalid_0's l2: 2.33788\n",
            "[338]\tvalid_0's l2: 2.33786\n",
            "[339]\tvalid_0's l2: 2.33768\n",
            "[340]\tvalid_0's l2: 2.33726\n",
            "[341]\tvalid_0's l2: 2.33613\n",
            "[342]\tvalid_0's l2: 2.33513\n",
            "[343]\tvalid_0's l2: 2.33415\n",
            "[344]\tvalid_0's l2: 2.33436\n",
            "[345]\tvalid_0's l2: 2.33338\n",
            "[346]\tvalid_0's l2: 2.3336\n",
            "[347]\tvalid_0's l2: 2.33443\n",
            "[348]\tvalid_0's l2: 2.33584\n",
            "[349]\tvalid_0's l2: 2.33662\n",
            "[350]\tvalid_0's l2: 2.33606\n",
            "[351]\tvalid_0's l2: 2.33634\n",
            "[352]\tvalid_0's l2: 2.33569\n",
            "[353]\tvalid_0's l2: 2.33604\n",
            "[354]\tvalid_0's l2: 2.33526\n",
            "[355]\tvalid_0's l2: 2.33525\n",
            "[356]\tvalid_0's l2: 2.33557\n",
            "[357]\tvalid_0's l2: 2.33526\n",
            "[358]\tvalid_0's l2: 2.33385\n",
            "[359]\tvalid_0's l2: 2.33409\n",
            "[360]\tvalid_0's l2: 2.33457\n",
            "[361]\tvalid_0's l2: 2.33464\n",
            "[362]\tvalid_0's l2: 2.33317\n",
            "[363]\tvalid_0's l2: 2.33305\n",
            "[364]\tvalid_0's l2: 2.33231\n",
            "[365]\tvalid_0's l2: 2.33099\n",
            "[366]\tvalid_0's l2: 2.32988\n",
            "[367]\tvalid_0's l2: 2.33011\n",
            "[368]\tvalid_0's l2: 2.33004\n",
            "[369]\tvalid_0's l2: 2.32968\n",
            "[370]\tvalid_0's l2: 2.32932\n",
            "[371]\tvalid_0's l2: 2.32927\n",
            "[372]\tvalid_0's l2: 2.32871\n",
            "[373]\tvalid_0's l2: 2.32628\n",
            "[374]\tvalid_0's l2: 2.32446\n",
            "[375]\tvalid_0's l2: 2.32476\n",
            "[376]\tvalid_0's l2: 2.32444\n",
            "[377]\tvalid_0's l2: 2.32228\n",
            "[378]\tvalid_0's l2: 2.32218\n",
            "[379]\tvalid_0's l2: 2.3216\n",
            "[380]\tvalid_0's l2: 2.32142\n",
            "[381]\tvalid_0's l2: 2.32149\n",
            "[382]\tvalid_0's l2: 2.3206\n",
            "[383]\tvalid_0's l2: 2.32055\n",
            "[384]\tvalid_0's l2: 2.32071\n",
            "[385]\tvalid_0's l2: 2.32088\n",
            "[386]\tvalid_0's l2: 2.32129\n",
            "[387]\tvalid_0's l2: 2.32175\n",
            "[388]\tvalid_0's l2: 2.32277\n",
            "[389]\tvalid_0's l2: 2.32179\n",
            "[390]\tvalid_0's l2: 2.32104\n",
            "[391]\tvalid_0's l2: 2.3213\n",
            "[392]\tvalid_0's l2: 2.32147\n",
            "[393]\tvalid_0's l2: 2.3206\n",
            "[394]\tvalid_0's l2: 2.31973\n",
            "[395]\tvalid_0's l2: 2.31815\n",
            "[396]\tvalid_0's l2: 2.31731\n",
            "[397]\tvalid_0's l2: 2.31803\n",
            "[398]\tvalid_0's l2: 2.31877\n",
            "[399]\tvalid_0's l2: 2.31938\n",
            "[400]\tvalid_0's l2: 2.31917\n",
            "[401]\tvalid_0's l2: 2.31993\n",
            "[402]\tvalid_0's l2: 2.31986\n",
            "[403]\tvalid_0's l2: 2.31895\n",
            "[404]\tvalid_0's l2: 2.31844\n",
            "[405]\tvalid_0's l2: 2.31824\n",
            "[406]\tvalid_0's l2: 2.31776\n",
            "[407]\tvalid_0's l2: 2.31692\n",
            "[408]\tvalid_0's l2: 2.31563\n",
            "[409]\tvalid_0's l2: 2.31557\n",
            "[410]\tvalid_0's l2: 2.31649\n",
            "[411]\tvalid_0's l2: 2.31621\n",
            "[412]\tvalid_0's l2: 2.31576\n",
            "[413]\tvalid_0's l2: 2.31637\n",
            "[414]\tvalid_0's l2: 2.31559\n",
            "[415]\tvalid_0's l2: 2.315\n",
            "[416]\tvalid_0's l2: 2.31413\n",
            "[417]\tvalid_0's l2: 2.31409\n",
            "[418]\tvalid_0's l2: 2.31338\n",
            "[419]\tvalid_0's l2: 2.31358\n",
            "[420]\tvalid_0's l2: 2.31303\n",
            "[421]\tvalid_0's l2: 2.31387\n",
            "[422]\tvalid_0's l2: 2.3132\n",
            "[423]\tvalid_0's l2: 2.31428\n",
            "[424]\tvalid_0's l2: 2.31469\n",
            "[425]\tvalid_0's l2: 2.31435\n",
            "[426]\tvalid_0's l2: 2.31304\n",
            "[427]\tvalid_0's l2: 2.31236\n",
            "[428]\tvalid_0's l2: 2.31151\n",
            "[429]\tvalid_0's l2: 2.31059\n",
            "[430]\tvalid_0's l2: 2.31017\n",
            "[431]\tvalid_0's l2: 2.30922\n",
            "[432]\tvalid_0's l2: 2.30957\n",
            "[433]\tvalid_0's l2: 2.30971\n",
            "[434]\tvalid_0's l2: 2.30958\n",
            "[435]\tvalid_0's l2: 2.30993\n",
            "[436]\tvalid_0's l2: 2.31029\n",
            "[437]\tvalid_0's l2: 2.30965\n",
            "[438]\tvalid_0's l2: 2.30989\n",
            "[439]\tvalid_0's l2: 2.30967\n",
            "[440]\tvalid_0's l2: 2.30892\n",
            "[441]\tvalid_0's l2: 2.30721\n",
            "[442]\tvalid_0's l2: 2.30693\n",
            "[443]\tvalid_0's l2: 2.30665\n",
            "[444]\tvalid_0's l2: 2.30598\n",
            "[445]\tvalid_0's l2: 2.3062\n",
            "[446]\tvalid_0's l2: 2.30616\n",
            "[447]\tvalid_0's l2: 2.30673\n",
            "[448]\tvalid_0's l2: 2.3057\n",
            "[449]\tvalid_0's l2: 2.30511\n",
            "[450]\tvalid_0's l2: 2.30406\n",
            "[451]\tvalid_0's l2: 2.30316\n",
            "[452]\tvalid_0's l2: 2.30187\n",
            "[453]\tvalid_0's l2: 2.30074\n",
            "[454]\tvalid_0's l2: 2.30099\n",
            "[455]\tvalid_0's l2: 2.30033\n",
            "[456]\tvalid_0's l2: 2.30016\n",
            "[457]\tvalid_0's l2: 2.29952\n",
            "[458]\tvalid_0's l2: 2.2994\n",
            "[459]\tvalid_0's l2: 2.29875\n",
            "[460]\tvalid_0's l2: 2.29851\n",
            "[461]\tvalid_0's l2: 2.29822\n",
            "[462]\tvalid_0's l2: 2.29757\n",
            "[463]\tvalid_0's l2: 2.29667\n",
            "[464]\tvalid_0's l2: 2.29637\n",
            "[465]\tvalid_0's l2: 2.29672\n",
            "[466]\tvalid_0's l2: 2.29581\n",
            "[467]\tvalid_0's l2: 2.29469\n",
            "[468]\tvalid_0's l2: 2.29448\n",
            "[469]\tvalid_0's l2: 2.2946\n",
            "[470]\tvalid_0's l2: 2.29465\n",
            "[471]\tvalid_0's l2: 2.29361\n",
            "[472]\tvalid_0's l2: 2.29243\n",
            "[473]\tvalid_0's l2: 2.29185\n",
            "[474]\tvalid_0's l2: 2.29158\n",
            "[475]\tvalid_0's l2: 2.29164\n",
            "[476]\tvalid_0's l2: 2.29136\n",
            "[477]\tvalid_0's l2: 2.29158\n",
            "[478]\tvalid_0's l2: 2.29148\n",
            "[479]\tvalid_0's l2: 2.29174\n",
            "[480]\tvalid_0's l2: 2.29097\n",
            "[481]\tvalid_0's l2: 2.29038\n",
            "[482]\tvalid_0's l2: 2.29013\n",
            "[483]\tvalid_0's l2: 2.28814\n",
            "[484]\tvalid_0's l2: 2.28917\n",
            "[485]\tvalid_0's l2: 2.28939\n",
            "[486]\tvalid_0's l2: 2.28953\n",
            "[487]\tvalid_0's l2: 2.28947\n",
            "[488]\tvalid_0's l2: 2.28883\n",
            "[489]\tvalid_0's l2: 2.28872\n",
            "[490]\tvalid_0's l2: 2.28924\n",
            "[491]\tvalid_0's l2: 2.28825\n",
            "[492]\tvalid_0's l2: 2.28927\n",
            "[493]\tvalid_0's l2: 2.28788\n",
            "[494]\tvalid_0's l2: 2.28813\n",
            "[495]\tvalid_0's l2: 2.28795\n",
            "[496]\tvalid_0's l2: 2.28763\n",
            "[497]\tvalid_0's l2: 2.28672\n",
            "[498]\tvalid_0's l2: 2.28768\n",
            "[499]\tvalid_0's l2: 2.28726\n",
            "[500]\tvalid_0's l2: 2.28866\n",
            "[501]\tvalid_0's l2: 2.2883\n",
            "[502]\tvalid_0's l2: 2.28826\n",
            "[503]\tvalid_0's l2: 2.28845\n",
            "[504]\tvalid_0's l2: 2.28848\n",
            "[505]\tvalid_0's l2: 2.28853\n",
            "[506]\tvalid_0's l2: 2.28803\n",
            "[507]\tvalid_0's l2: 2.28813\n",
            "[508]\tvalid_0's l2: 2.28832\n",
            "[509]\tvalid_0's l2: 2.28717\n",
            "[510]\tvalid_0's l2: 2.28677\n",
            "[511]\tvalid_0's l2: 2.28663\n",
            "[512]\tvalid_0's l2: 2.2865\n",
            "[513]\tvalid_0's l2: 2.28622\n",
            "[514]\tvalid_0's l2: 2.28552\n",
            "[515]\tvalid_0's l2: 2.28512\n",
            "[516]\tvalid_0's l2: 2.2851\n",
            "[517]\tvalid_0's l2: 2.28508\n",
            "[518]\tvalid_0's l2: 2.28394\n",
            "[519]\tvalid_0's l2: 2.2839\n",
            "[520]\tvalid_0's l2: 2.28384\n",
            "[521]\tvalid_0's l2: 2.2831\n",
            "[522]\tvalid_0's l2: 2.28348\n",
            "[523]\tvalid_0's l2: 2.28312\n",
            "[524]\tvalid_0's l2: 2.28337\n",
            "[525]\tvalid_0's l2: 2.28346\n",
            "[526]\tvalid_0's l2: 2.28337\n",
            "[527]\tvalid_0's l2: 2.28359\n",
            "[528]\tvalid_0's l2: 2.28369\n",
            "[529]\tvalid_0's l2: 2.28327\n",
            "[530]\tvalid_0's l2: 2.28353\n",
            "[531]\tvalid_0's l2: 2.28328\n",
            "[532]\tvalid_0's l2: 2.28212\n",
            "[533]\tvalid_0's l2: 2.2819\n",
            "[534]\tvalid_0's l2: 2.28184\n",
            "[535]\tvalid_0's l2: 2.28193\n",
            "[536]\tvalid_0's l2: 2.28146\n",
            "[537]\tvalid_0's l2: 2.28091\n",
            "[538]\tvalid_0's l2: 2.28119\n",
            "[539]\tvalid_0's l2: 2.28073\n",
            "[540]\tvalid_0's l2: 2.28035\n",
            "[541]\tvalid_0's l2: 2.28047\n",
            "[542]\tvalid_0's l2: 2.27999\n",
            "[543]\tvalid_0's l2: 2.27957\n",
            "[544]\tvalid_0's l2: 2.27989\n",
            "[545]\tvalid_0's l2: 2.28055\n",
            "[546]\tvalid_0's l2: 2.28081\n",
            "[547]\tvalid_0's l2: 2.2811\n",
            "[548]\tvalid_0's l2: 2.28177\n",
            "[549]\tvalid_0's l2: 2.28109\n",
            "[550]\tvalid_0's l2: 2.28075\n",
            "[551]\tvalid_0's l2: 2.27983\n",
            "[552]\tvalid_0's l2: 2.2799\n",
            "[553]\tvalid_0's l2: 2.27987\n",
            "[554]\tvalid_0's l2: 2.27976\n",
            "[555]\tvalid_0's l2: 2.27889\n",
            "[556]\tvalid_0's l2: 2.27826\n",
            "[557]\tvalid_0's l2: 2.27841\n",
            "[558]\tvalid_0's l2: 2.27803\n",
            "[559]\tvalid_0's l2: 2.27795\n",
            "[560]\tvalid_0's l2: 2.27809\n",
            "[561]\tvalid_0's l2: 2.27795\n",
            "[562]\tvalid_0's l2: 2.27774\n",
            "[563]\tvalid_0's l2: 2.27781\n",
            "[564]\tvalid_0's l2: 2.27775\n",
            "[565]\tvalid_0's l2: 2.27844\n",
            "[566]\tvalid_0's l2: 2.27803\n",
            "[567]\tvalid_0's l2: 2.27744\n",
            "[568]\tvalid_0's l2: 2.27746\n",
            "[569]\tvalid_0's l2: 2.27699\n",
            "[570]\tvalid_0's l2: 2.27686\n",
            "[571]\tvalid_0's l2: 2.27678\n",
            "[572]\tvalid_0's l2: 2.27644\n",
            "[573]\tvalid_0's l2: 2.27676\n",
            "[574]\tvalid_0's l2: 2.27639\n",
            "[575]\tvalid_0's l2: 2.2758\n",
            "[576]\tvalid_0's l2: 2.27518\n",
            "[577]\tvalid_0's l2: 2.27426\n",
            "[578]\tvalid_0's l2: 2.27462\n",
            "[579]\tvalid_0's l2: 2.27482\n",
            "[580]\tvalid_0's l2: 2.27491\n",
            "[581]\tvalid_0's l2: 2.27517\n",
            "[582]\tvalid_0's l2: 2.2746\n",
            "[583]\tvalid_0's l2: 2.27417\n",
            "[584]\tvalid_0's l2: 2.27359\n",
            "[585]\tvalid_0's l2: 2.27324\n",
            "[586]\tvalid_0's l2: 2.27237\n",
            "[587]\tvalid_0's l2: 2.2724\n",
            "[588]\tvalid_0's l2: 2.2716\n",
            "[589]\tvalid_0's l2: 2.27132\n",
            "[590]\tvalid_0's l2: 2.27119\n",
            "[591]\tvalid_0's l2: 2.27102\n",
            "[592]\tvalid_0's l2: 2.27076\n",
            "[593]\tvalid_0's l2: 2.27089\n",
            "[594]\tvalid_0's l2: 2.27059\n",
            "[595]\tvalid_0's l2: 2.27018\n",
            "[596]\tvalid_0's l2: 2.2704\n",
            "[597]\tvalid_0's l2: 2.27022\n",
            "[598]\tvalid_0's l2: 2.27024\n",
            "[599]\tvalid_0's l2: 2.26961\n",
            "[600]\tvalid_0's l2: 2.26913\n",
            "[601]\tvalid_0's l2: 2.26924\n",
            "[602]\tvalid_0's l2: 2.26846\n",
            "[603]\tvalid_0's l2: 2.26757\n",
            "[604]\tvalid_0's l2: 2.26699\n",
            "[605]\tvalid_0's l2: 2.26679\n",
            "[606]\tvalid_0's l2: 2.26597\n",
            "[607]\tvalid_0's l2: 2.26592\n",
            "[608]\tvalid_0's l2: 2.26662\n",
            "[609]\tvalid_0's l2: 2.26641\n",
            "[610]\tvalid_0's l2: 2.26638\n",
            "[611]\tvalid_0's l2: 2.26694\n",
            "[612]\tvalid_0's l2: 2.26736\n",
            "[613]\tvalid_0's l2: 2.26768\n",
            "[614]\tvalid_0's l2: 2.26798\n",
            "[615]\tvalid_0's l2: 2.26734\n",
            "[616]\tvalid_0's l2: 2.26638\n",
            "[617]\tvalid_0's l2: 2.2666\n",
            "[618]\tvalid_0's l2: 2.26629\n",
            "[619]\tvalid_0's l2: 2.26649\n",
            "[620]\tvalid_0's l2: 2.26657\n",
            "[621]\tvalid_0's l2: 2.26562\n",
            "[622]\tvalid_0's l2: 2.26499\n",
            "[623]\tvalid_0's l2: 2.26544\n",
            "[624]\tvalid_0's l2: 2.26637\n",
            "[625]\tvalid_0's l2: 2.26621\n",
            "[626]\tvalid_0's l2: 2.26648\n",
            "[627]\tvalid_0's l2: 2.26607\n",
            "[628]\tvalid_0's l2: 2.26619\n",
            "[629]\tvalid_0's l2: 2.26579\n",
            "[630]\tvalid_0's l2: 2.2661\n",
            "[631]\tvalid_0's l2: 2.26584\n",
            "[632]\tvalid_0's l2: 2.2647\n",
            "[633]\tvalid_0's l2: 2.26472\n",
            "[634]\tvalid_0's l2: 2.26464\n",
            "[635]\tvalid_0's l2: 2.26413\n",
            "[636]\tvalid_0's l2: 2.26347\n",
            "[637]\tvalid_0's l2: 2.26357\n",
            "[638]\tvalid_0's l2: 2.26331\n",
            "[639]\tvalid_0's l2: 2.2632\n",
            "[640]\tvalid_0's l2: 2.26284\n",
            "[641]\tvalid_0's l2: 2.26262\n",
            "[642]\tvalid_0's l2: 2.26223\n",
            "[643]\tvalid_0's l2: 2.26272\n",
            "[644]\tvalid_0's l2: 2.26211\n",
            "[645]\tvalid_0's l2: 2.26169\n",
            "[646]\tvalid_0's l2: 2.26113\n",
            "[647]\tvalid_0's l2: 2.26122\n",
            "[648]\tvalid_0's l2: 2.26127\n",
            "[649]\tvalid_0's l2: 2.26112\n",
            "[650]\tvalid_0's l2: 2.2611\n",
            "[651]\tvalid_0's l2: 2.26155\n",
            "[652]\tvalid_0's l2: 2.26126\n",
            "[653]\tvalid_0's l2: 2.26102\n",
            "[654]\tvalid_0's l2: 2.26186\n",
            "[655]\tvalid_0's l2: 2.26183\n",
            "[656]\tvalid_0's l2: 2.26185\n",
            "[657]\tvalid_0's l2: 2.26256\n",
            "[658]\tvalid_0's l2: 2.26302\n",
            "[659]\tvalid_0's l2: 2.26254\n",
            "[660]\tvalid_0's l2: 2.26245\n",
            "[661]\tvalid_0's l2: 2.26228\n",
            "[662]\tvalid_0's l2: 2.26275\n",
            "[663]\tvalid_0's l2: 2.26301\n",
            "[664]\tvalid_0's l2: 2.26326\n",
            "[665]\tvalid_0's l2: 2.26349\n",
            "[666]\tvalid_0's l2: 2.26382\n",
            "[667]\tvalid_0's l2: 2.26475\n",
            "[668]\tvalid_0's l2: 2.26488\n",
            "[669]\tvalid_0's l2: 2.26491\n",
            "[670]\tvalid_0's l2: 2.26569\n",
            "[671]\tvalid_0's l2: 2.26575\n",
            "[672]\tvalid_0's l2: 2.26527\n",
            "[673]\tvalid_0's l2: 2.26532\n",
            "[674]\tvalid_0's l2: 2.26495\n",
            "[675]\tvalid_0's l2: 2.26525\n",
            "[676]\tvalid_0's l2: 2.26501\n",
            "[677]\tvalid_0's l2: 2.26557\n",
            "[678]\tvalid_0's l2: 2.26536\n",
            "Early stopping, best iteration is:\n",
            "[653]\tvalid_0's l2: 2.26102\n",
            "CPU times: user 14.4 s, sys: 275 ms, total: 14.7 s\n",
            "Wall time: 7.73 s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FAjUjctV-IDU",
        "colab_type": "text"
      },
      "source": [
        "# Comparing Results"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Rbf7Rb5k9m5I",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 255
        },
        "outputId": "e62acba1-fee4-4c71-8405-3578a7001d2b"
      },
      "source": [
        "print('CatBoost')\n",
        "regression_report(catboost_model, x_train, y_train)\n",
        "regression_report(catboost_model, x_test, y_test)\n",
        "\n",
        "print('LightGBM')\n",
        "regression_report(light_model, x_train, y_train)\n",
        "regression_report(light_model, x_test, y_test)"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CatBoost\n",
            "Rsq = 97.8398%\n",
            "MSE = 0.7105\n",
            "\n",
            "Rsq = 95.3538%\n",
            "MSE = 1.5520\n",
            "\n",
            "LightGBM\n",
            "Rsq = 99.3146%\n",
            "MSE = 0.2254\n",
            "\n",
            "Rsq = 92.7046%\n",
            "MSE = 2.4369\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tjqgIpFw_ZlJ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 702
        },
        "outputId": "0cb73b12-d5f7-4ff6-880e-d1f5dbba162a"
      },
      "source": [
        "ψ_catboost = catboost_model.predict(x1) - catboost_model.predict(x0)\n",
        "ψ_catboost_dr = catboost_dr_model.predict(x1) - catboost_dr_model.predict(x0)\n",
        "ψ_lightgbm = light_model.predict(x1) - light_model.predict(x0)\n",
        "ψ_lightgbm_dr = light_dr_model.predict(x1) - light_dr_model.predict(x0)\n",
        "\n",
        "print('CatBoost')\n",
        "causal_report(ψ_catboost, true_effect)\n",
        "\n",
        "print('Catboost DR')\n",
        "causal_report(ψ_catboost_dr, true_effect)\n",
        "\n",
        "print('LightGBM')\n",
        "causal_report(ψ_lightgbm, true_effect)\n",
        "\n",
        "print('LightGBM DR')\n",
        "causal_report(ψ_lightgbm_dr, true_effect)\n",
        "\n",
        "\n",
        "graph.figure(figsize=(15, 5))\n",
        "sns.distplot(ψ_catboost, hist=False, label='CatBoost')\n",
        "sns.distplot(ψ_catboost_dr, hist=False, label='CatBoost DR')\n",
        "sns.distplot(ψ_lightgbm, hist=False, label='LightGBM')\n",
        "sns.distplot(ψ_lightgbm_dr, hist=False, label='LightGBM DR')\n",
        "graph.axvline(true_effect, color='black')\n",
        "graph.xlabel('Ψ')\n",
        "graph.legend()\n",
        "graph.show()"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CatBoost\n",
            "Mean Ψ = 1.0372\n",
            "Mean Error of Ψ\n",
            "    0.1672\n",
            "\n",
            "Catboost DR\n",
            "Mean Ψ = 1.0016\n",
            "Mean Error of Ψ\n",
            "    0.1674\n",
            "\n",
            "LightGBM\n",
            "Mean Ψ = 0.9886\n",
            "Mean Error of Ψ\n",
            "    0.3960\n",
            "\n",
            "LightGBM DR\n",
            "Mean Ψ = 0.9711\n",
            "Mean Error of Ψ\n",
            "    0.3929\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ANmD1Gf2B0Ic",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 948
        },
        "outputId": "4e21fd24-613a-429e-e9ba-d3636822fbf4"
      },
      "source": [
        "cate_model = sm.WLS(\n",
        "    ψ_catboost_dr, \n",
        "    sm.add_constant(x.drop(columns='treatment')),\n",
        "    weights=g\n",
        ").fit()\n",
        "\n",
        "display(cate_model.summary())"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table class=\"simpletable\">\n",
              "<caption>WLS Regression Results</caption>\n",
              "<tr>\n",
              "  <th>Dep. Variable:</th>            <td>y</td>        <th>  R-squared:         </th> <td>   0.079</td> \n",
              "</tr>\n",
              "<tr>\n",
              "  <th>Model:</th>                   <td>WLS</td>       <th>  Adj. R-squared:    </th> <td>   0.076</td> \n",
              "</tr>\n",
              "<tr>\n",
              "  <th>Method:</th>             <td>Least Squares</td>  <th>  F-statistic:       </th> <td>   34.04</td> \n",
              "</tr>\n",
              "<tr>\n",
              "  <th>Date:</th>             <td>Mon, 22 Jun 2020</td> <th>  Prob (F-statistic):</th> <td>2.47e-156</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>Time:</th>                 <td>04:30:11</td>     <th>  Log-Likelihood:    </th> <td>  548.70</td> \n",
              "</tr>\n",
              "<tr>\n",
              "  <th>No. Observations:</th>      <td> 10000</td>      <th>  AIC:               </th> <td>  -1045.</td> \n",
              "</tr>\n",
              "<tr>\n",
              "  <th>Df Residuals:</th>          <td>  9974</td>      <th>  BIC:               </th> <td>  -857.9</td> \n",
              "</tr>\n",
              "<tr>\n",
              "  <th>Df Model:</th>              <td>    25</td>      <th>                     </th>     <td> </td>    \n",
              "</tr>\n",
              "<tr>\n",
              "  <th>Covariance Type:</th>      <td>nonrobust</td>    <th>                     </th>     <td> </td>    \n",
              "</tr>\n",
              "</table>\n",
              "<table class=\"simpletable\">\n",
              "<tr>\n",
              "    <td></td>       <th>coef</th>     <th>std err</th>      <th>t</th>      <th>P>|t|</th>  <th>[0.025</th>    <th>0.975]</th>  \n",
              "</tr>\n",
              "<tr>\n",
              "  <th>const</th> <td>    0.9919</td> <td>    0.002</td> <td>  410.867</td> <td> 0.000</td> <td>    0.987</td> <td>    0.997</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>0</th>     <td>    0.0086</td> <td>    0.002</td> <td>    3.682</td> <td> 0.000</td> <td>    0.004</td> <td>    0.013</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>1</th>     <td>    0.0079</td> <td>    0.002</td> <td>    3.573</td> <td> 0.000</td> <td>    0.004</td> <td>    0.012</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>2</th>     <td>    0.0053</td> <td>    0.002</td> <td>    2.513</td> <td> 0.012</td> <td>    0.001</td> <td>    0.010</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>3</th>     <td>   -0.0107</td> <td>    0.002</td> <td>   -5.030</td> <td> 0.000</td> <td>   -0.015</td> <td>   -0.007</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>4</th>     <td>    0.0006</td> <td>    0.002</td> <td>    0.257</td> <td> 0.797</td> <td>   -0.004</td> <td>    0.005</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>5</th>     <td>   -0.0034</td> <td>    0.002</td> <td>   -1.611</td> <td> 0.107</td> <td>   -0.008</td> <td>    0.001</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>6</th>     <td>   -0.0124</td> <td>    0.002</td> <td>   -5.729</td> <td> 0.000</td> <td>   -0.017</td> <td>   -0.008</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>7</th>     <td>    0.0118</td> <td>    0.002</td> <td>    5.520</td> <td> 0.000</td> <td>    0.008</td> <td>    0.016</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>8</th>     <td>    0.0079</td> <td>    0.002</td> <td>    3.699</td> <td> 0.000</td> <td>    0.004</td> <td>    0.012</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>9</th>     <td>   -0.0185</td> <td>    0.002</td> <td>   -8.748</td> <td> 0.000</td> <td>   -0.023</td> <td>   -0.014</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>10</th>    <td>   -0.0055</td> <td>    0.002</td> <td>   -2.588</td> <td> 0.010</td> <td>   -0.010</td> <td>   -0.001</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>11</th>    <td>    0.0169</td> <td>    0.002</td> <td>    7.971</td> <td> 0.000</td> <td>    0.013</td> <td>    0.021</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>12</th>    <td>   -0.0035</td> <td>    0.002</td> <td>   -1.646</td> <td> 0.100</td> <td>   -0.008</td> <td>    0.001</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>13</th>    <td>    0.0124</td> <td>    0.002</td> <td>    5.851</td> <td> 0.000</td> <td>    0.008</td> <td>    0.017</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>14</th>    <td>   -0.0074</td> <td>    0.002</td> <td>   -3.557</td> <td> 0.000</td> <td>   -0.011</td> <td>   -0.003</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>15</th>    <td>    0.0247</td> <td>    0.002</td> <td>   11.759</td> <td> 0.000</td> <td>    0.021</td> <td>    0.029</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>16</th>    <td>    0.0004</td> <td>    0.002</td> <td>    0.206</td> <td> 0.837</td> <td>   -0.004</td> <td>    0.005</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>17</th>    <td>   -0.0026</td> <td>    0.002</td> <td>   -1.252</td> <td> 0.211</td> <td>   -0.007</td> <td>    0.001</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>18</th>    <td>    0.0039</td> <td>    0.002</td> <td>    1.848</td> <td> 0.065</td> <td>   -0.000</td> <td>    0.008</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>19</th>    <td>   -0.0102</td> <td>    0.002</td> <td>   -4.861</td> <td> 0.000</td> <td>   -0.014</td> <td>   -0.006</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>20</th>    <td>   -0.0027</td> <td>    0.002</td> <td>   -1.268</td> <td> 0.205</td> <td>   -0.007</td> <td>    0.001</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>21</th>    <td>   -0.0389</td> <td>    0.002</td> <td>  -18.279</td> <td> 0.000</td> <td>   -0.043</td> <td>   -0.035</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>22</th>    <td>    0.0054</td> <td>    0.002</td> <td>    2.542</td> <td> 0.011</td> <td>    0.001</td> <td>    0.010</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>23</th>    <td>   -0.0012</td> <td>    0.002</td> <td>   -0.597</td> <td> 0.550</td> <td>   -0.005</td> <td>    0.003</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>24</th>    <td>   -0.0043</td> <td>    0.002</td> <td>   -2.054</td> <td> 0.040</td> <td>   -0.008</td> <td>   -0.000</td>\n",
              "</tr>\n",
              "</table>\n",
              "<table class=\"simpletable\">\n",
              "<tr>\n",
              "  <th>Omnibus:</th>       <td>1082.952</td> <th>  Durbin-Watson:     </th> <td>   1.965</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>Prob(Omnibus):</th>  <td> 0.000</td>  <th>  Jarque-Bera (JB):  </th> <td>7218.348</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>Skew:</th>           <td>-0.296</td>  <th>  Prob(JB):          </th> <td>    0.00</td>\n",
              "</tr>\n",
              "<tr>\n",
              "  <th>Kurtosis:</th>       <td> 7.120</td>  <th>  Cond. No.          </th> <td>    1.69</td>\n",
              "</tr>\n",
              "</table><br/><br/>Warnings:<br/>[1] Standard Errors assume that the covariance matrix of the errors is correctly specified."
            ],
            "text/plain": [
              "<class 'statsmodels.iolib.summary.Summary'>\n",
              "\"\"\"\n",
              "                            WLS Regression Results                            \n",
              "==============================================================================\n",
              "Dep. Variable:                      y   R-squared:                       0.079\n",
              "Model:                            WLS   Adj. R-squared:                  0.076\n",
              "Method:                 Least Squares   F-statistic:                     34.04\n",
              "Date:                Mon, 22 Jun 2020   Prob (F-statistic):          2.47e-156\n",
              "Time:                        04:30:11   Log-Likelihood:                 548.70\n",
              "No. Observations:               10000   AIC:                            -1045.\n",
              "Df Residuals:                    9974   BIC:                            -857.9\n",
              "Df Model:                          25                                         \n",
              "Covariance Type:            nonrobust                                         \n",
              "==============================================================================\n",
              "                 coef    std err          t      P>|t|      [0.025      0.975]\n",
              "------------------------------------------------------------------------------\n",
              "const          0.9919      0.002    410.867      0.000       0.987       0.997\n",
              "0              0.0086      0.002      3.682      0.000       0.004       0.013\n",
              "1              0.0079      0.002      3.573      0.000       0.004       0.012\n",
              "2              0.0053      0.002      2.513      0.012       0.001       0.010\n",
              "3             -0.0107      0.002     -5.030      0.000      -0.015      -0.007\n",
              "4              0.0006      0.002      0.257      0.797      -0.004       0.005\n",
              "5             -0.0034      0.002     -1.611      0.107      -0.008       0.001\n",
              "6             -0.0124      0.002     -5.729      0.000      -0.017      -0.008\n",
              "7              0.0118      0.002      5.520      0.000       0.008       0.016\n",
              "8              0.0079      0.002      3.699      0.000       0.004       0.012\n",
              "9             -0.0185      0.002     -8.748      0.000      -0.023      -0.014\n",
              "10            -0.0055      0.002     -2.588      0.010      -0.010      -0.001\n",
              "11             0.0169      0.002      7.971      0.000       0.013       0.021\n",
              "12            -0.0035      0.002     -1.646      0.100      -0.008       0.001\n",
              "13             0.0124      0.002      5.851      0.000       0.008       0.017\n",
              "14            -0.0074      0.002     -3.557      0.000      -0.011      -0.003\n",
              "15             0.0247      0.002     11.759      0.000       0.021       0.029\n",
              "16             0.0004      0.002      0.206      0.837      -0.004       0.005\n",
              "17            -0.0026      0.002     -1.252      0.211      -0.007       0.001\n",
              "18             0.0039      0.002      1.848      0.065      -0.000       0.008\n",
              "19            -0.0102      0.002     -4.861      0.000      -0.014      -0.006\n",
              "20            -0.0027      0.002     -1.268      0.205      -0.007       0.001\n",
              "21            -0.0389      0.002    -18.279      0.000      -0.043      -0.035\n",
              "22             0.0054      0.002      2.542      0.011       0.001       0.010\n",
              "23            -0.0012      0.002     -0.597      0.550      -0.005       0.003\n",
              "24            -0.0043      0.002     -2.054      0.040      -0.008      -0.000\n",
              "==============================================================================\n",
              "Omnibus:                     1082.952   Durbin-Watson:                   1.965\n",
              "Prob(Omnibus):                  0.000   Jarque-Bera (JB):             7218.348\n",
              "Skew:                          -0.296   Prob(JB):                         0.00\n",
              "Kurtosis:                       7.120   Cond. No.                         1.69\n",
              "==============================================================================\n",
              "\n",
              "Warnings:\n",
              "[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.\n",
              "\"\"\""
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0-UPlhNlOyPu",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 787
        },
        "outputId": "89d0c970-8c73-43d5-c3f1-3a023b04626b"
      },
      "source": [
        "print('CatBoost')\n",
        "causal_report(ψ_catboost, true_effect)\n",
        "\n",
        "print('Catboost DR')\n",
        "causal_report(ψ_catboost_dr, true_effect)\n",
        "\n",
        "print('CatBoost DR with WLS-Ψ')\n",
        "causal_report(cate_model.fittedvalues, true_effect)\n",
        "\n",
        "print('LightGBM')\n",
        "causal_report(ψ_lightgbm, true_effect)\n",
        "\n",
        "print('LightGBM DR')\n",
        "causal_report(ψ_lightgbm_dr, true_effect)\n",
        "\n",
        "\n",
        "graph.figure(figsize=(15, 5))\n",
        "sns.distplot(ψ_catboost, hist=False, label='CatBoost')\n",
        "sns.distplot(ψ_catboost_dr, hist=False, label='CatBoost DR')\n",
        "sns.distplot(ψ_lightgbm, hist=False, label='LightGBM')\n",
        "sns.distplot(ψ_lightgbm_dr, hist=False, label='LightGBM DR')\n",
        "sns.distplot(cate_model.fittedvalues, hist=False, label='CatBoost DR with Polish')\n",
        "graph.axvline(true_effect, color='black')\n",
        "graph.xlabel('Ψ')\n",
        "graph.legend()\n",
        "graph.show()"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CatBoost\n",
            "Mean Ψ = 1.0372\n",
            "Mean Error of Ψ\n",
            "    0.1672\n",
            "\n",
            "Catboost DR\n",
            "Mean Ψ = 1.0016\n",
            "Mean Error of Ψ\n",
            "    0.1674\n",
            "\n",
            "CatBoost DR with WLS-Ψ\n",
            "Mean Ψ = 0.9932\n",
            "Mean Error of Ψ\n",
            "    0.0499\n",
            "\n",
            "LightGBM\n",
            "Mean Ψ = 0.9886\n",
            "Mean Error of Ψ\n",
            "    0.3960\n",
            "\n",
            "LightGBM DR\n",
            "Mean Ψ = 0.9711\n",
            "Mean Error of Ψ\n",
            "    0.3929\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}