nb5_ml-windfarm.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Notebook 5 - Building ML models for the windfarm dataset\n",
    "\n",
    "Now let's get cracking. Here we will build different models, just as we have done for the first dataset. We will also try to learn something about the dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import pickle\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.metrics import r2_score\n",
    "from keras.models import Sequential\n",
    "from keras.layers import *\n",
    "from keras.optimizers import SGD, Adam"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "DATADIR = './dataset2/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "with open('./tmp/wind_ds_clean.pkl', 'rb') as f:\n",
    "    X_train, Xobs_train, y_train, X_valid, Xobs_valid, y_valid, X_test, Xobs_test = pickle.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "with open('./tmp/test_times.pkl', 'rb') as f:\n",
    "    test_times = pickle.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((2507, 210), (2507, 24), (2507, 33), (199, 210))"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape, Xobs_train.shape, y_train.shape, X_valid.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def mse(y_true, y_pred, axis=(0,1)): return ((y_true - y_pred)**2).mean(axis)\n",
    "def rmse(y_true, y_pred, axis=(0,1)): return np.sqrt(((y_true - y_pred)**2).mean(axis))\n",
    "def create_sub(preds, fn=None):\n",
    "    \"\"\"Sorry for the complicated submission format. But just use this function \"\"\"\n",
    "    time_string = [\n",
    "        f'{str(tt)[:13]}+{str(lt).zfill(2)}h' for tt, lt \n",
    "        in zip(np.repeat(test_times, 33), np.tile(np.arange(0, 33*3,3), len(test_times)))]\n",
    "    df =  pd.DataFrame({'Time': time_string, 'Expected': np.ravel(preds)})\n",
    "    if fn is not None: df.to_csv(DATADIR + fn, index=False)\n",
    "    return df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Linear regression\n",
    "\n",
    "Let's start out again with a simple linear regression model. It is always good to start with the dumbest model possible to create a benchmark."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Normalize the dataset\n",
    "\n",
    "Recall that for linear regression and neural networks we need to normalize our input data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Remember to use the same mean and std for all datasets\n",
    "mean, std = X_train.mean(0), X_train.std(0)\n",
    "X_train = (X_train - mean) / std\n",
    "X_valid = (X_valid - mean) / std\n",
    "X_test = (X_test - mean) / std"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x20c3c85ad30>]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAD8CAYAAAB+UHOxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nyJEjR45litwA5MiRI8cyxf8PPWDbDhCZnMQAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x20c3c85ae10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(X_train[-1])\n",
    "plt.plot(X_valid[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train the model\n",
    "\n",
    "Now we proceed as before."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearRegression(copy_X=True, fit_intercept=True, n_jobs=1, normalize=False)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr = LinearRegression()\n",
    "lr.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def print_scores(m, X_train=X_train, X_valid=X_valid):\n",
    "    print('Train R2 = ', m.score(X_train, y_train), \n",
    "          ', Valid R2 = ', m.score(X_valid, y_valid), ', Valid RMSE = ', \n",
    "          rmse(m.predict(X_valid), y_valid))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train R2 =  0.551317654454 , Valid R2 =  0.443831726675 , Valid RMSE =  0.16552\n"
     ]
    }
   ],
   "source": [
    "print_scores(lr)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Check the prediction against lead time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def plot_rmse_lead(y_true, y_pred, label=None):\n",
    "    plt.plot(np.arange(0, y_true.shape[1]*3, 3), rmse(y_true, y_pred, 0), label=label)\n",
    "    plt.xlabel('Lead time'); plt.ylabel('RMSE'); \n",
    "    if label is not None: plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAY8AAAEKCAYAAADq59mMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x20c3cc3ac88>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_rmse_lead(y_train, lr.predict(X_train), 'lr train')\n",
    "plot_rmse_lead(y_valid, lr.predict(X_valid), 'lr valid')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that the error goes up with time, just as we would expect."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create submission"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df = create_sub(lr.predict(X_test), 'lr.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "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>Time</th>\n",
       "      <th>Expected</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2015-03-01T00+00h</td>\n",
       "      <td>0.151643</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2015-03-01T00+03h</td>\n",
       "      <td>0.072004</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2015-03-01T00+06h</td>\n",
       "      <td>-0.012968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2015-03-01T00+09h</td>\n",
       "      <td>0.064583</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2015-03-01T00+12h</td>\n",
       "      <td>0.159778</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                Time  Expected\n",
       "0  2015-03-01T00+00h  0.151643\n",
       "1  2015-03-01T00+03h  0.072004\n",
       "2  2015-03-01T00+06h -0.012968\n",
       "3  2015-03-01T00+09h  0.064583\n",
       "4  2015-03-01T00+12h  0.159778"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Random forest\n",
    "\n",
    "We already know that we can probably do better with a random forest. Se let's simply try our setting from the previous notebook."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "rf = RandomForestRegressor(n_estimators=40, n_jobs=-1, min_samples_leaf=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 5.24 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "RandomForestRegressor(bootstrap=True, criterion='mse', max_depth=None,\n",
       "           max_features='auto', max_leaf_nodes=None,\n",
       "           min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "           min_samples_leaf=10, min_samples_split=2,\n",
       "           min_weight_fraction_leaf=0.0, n_estimators=40, n_jobs=-1,\n",
       "           oob_score=False, random_state=None, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%time rf.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train R2 =  0.555403438363 , Valid R2 =  0.38464381991 , Valid RMSE =  0.174104323044\n"
     ]
    }
   ],
   "source": [
    "print_scores(rf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAY8AAAEKCAYAAADq59mMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x20c3cc88eb8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_rmse_lead(y_train, lr.predict(X_train), 'lr train')\n",
    "plot_rmse_lead(y_valid, lr.predict(X_valid), 'lr valid')\n",
    "plot_rmse_lead(y_train, rf.predict(X_train), 'rf train')\n",
    "plot_rmse_lead(y_valid, rf.predict(X_valid), 'rf valid')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Well, that is not very good. Even the training score is not better than for linear regression. This tells us that there is some underfitting going on. Let's try to increase the complexity of the forest."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### A more complex forest?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "rf = RandomForestRegressor(n_estimators=40, n_jobs=-1, min_samples_leaf=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 6.03 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "RandomForestRegressor(bootstrap=True, criterion='mse', max_depth=None,\n",
       "           max_features='auto', max_leaf_nodes=None,\n",
       "           min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "           min_samples_leaf=5, min_samples_split=2,\n",
       "           min_weight_fraction_leaf=0.0, n_estimators=40, n_jobs=-1,\n",
       "           oob_score=False, random_state=None, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%time rf.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train R2 =  0.658526635097 , Valid R2 =  0.39671102941 , Valid RMSE =  0.172388767367\n"
     ]
    }
   ],
   "source": [
    "print_scores(rf)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "That's a little better. But even after trying a range of parameter combinations I was not able to beat the linear regression score. Hmmm?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df = create_sub(rf.predict(X_test), 'rf.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As you can see, real machine learning problems are hard and a lot of the time is spent on wondering what the hell is going on. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Neural networks in sklearn\n",
    "\n",
    "Ok, so now let's start with a neural network. At first, let's actually try something different and use the sklearn MLP class. It's very unflexible but has some good defaults for simple problems like ours."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.neural_network import MLPRegressor\n",
    "nn = MLPRegressor(hidden_layer_sizes=[32, 32], activation='tanh', max_iter=200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 668 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "MLPRegressor(activation='tanh', alpha=0.0001, batch_size='auto', beta_1=0.9,\n",
       "       beta_2=0.999, early_stopping=False, epsilon=1e-08,\n",
       "       hidden_layer_sizes=[32, 32], learning_rate='constant',\n",
       "       learning_rate_init=0.001, max_iter=200, momentum=0.9,\n",
       "       nesterovs_momentum=True, power_t=0.5, random_state=None,\n",
       "       shuffle=True, solver='adam', tol=0.0001, validation_fraction=0.1,\n",
       "       verbose=False, warm_start=False)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%time nn.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train R2 =  0.466047148095 , Valid R2 =  0.368379158573 , Valid RMSE =  0.176390218211\n"
     ]
    }
   ],
   "source": [
    "print_scores(nn)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I noticed something interesting when trying out the NN above."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_scores, valid_scores = [], []\n",
    "train_rmses, valid_rmses = [], []\n",
    "nns = []\n",
    "for i in range(20):\n",
    "    nn = MLPRegressor(hidden_layer_sizes=[32, 32], activation='tanh', max_iter=200)\n",
    "    nn.fit(X_train, y_train)\n",
    "    train_scores.append(nn.score(X_train, y_train)); valid_scores.append(nn.score(X_valid, y_valid))\n",
    "    train_rmses.append(rmse(nn.predict(X_train), y_train)); valid_rmses.append(rmse(nn.predict(X_valid), y_valid))\n",
    "    nns.append(nn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x20c3da814e0>]"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAD8CAYAAACb4nSYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAIABJREFUeJzt3Xd4VGX2wPHvSQVCqAk1QACRjpTQ\nm1hRXFBBRSxYARVxV91Vf+uubd1de4NVsWBdKVZ0UVREukgooXeBhFASCCUE0ub9/fFOcAgpk2Ra\nMufzPDyZufPee0+GyZl73yrGGJRSSgWHEH8HoJRSync06SulVBDRpK+UUkFEk75SSgURTfpKKRVE\nNOkrpVQQ0aSvlFJBRJO+UkoFEU36SikVRML8HUBhMTExJj4+3t9hKKVUpbJy5cp0Y0xsaeUCLunH\nx8eTmJjo7zCUUqpSEZHd7pTT6h2llAoimvSVUiqIaNJXSqkgoklfKaWCiCZ9pZQKIpr0lVIqiGjS\nV0qpIOJW0heRoSKyRUS2i8jDJZQbJSJGRBKcz8NF5H0RWScim0TkEU8FrpQKLN+t30fy4Sx/h6FK\nUWrSF5FQYApwGdABuF5EOhRRLhqYBCx32XwNEGmM6Qz0AMaLSHzFw1ZKBZK3F+1kwkereHBWkr9D\nUaVw50q/F7DdGLPTGJMDTAdGFFHuKeBZ4JTLNgNEiUgYUB3IAY5VLGSlVCD5fFUK//jfJhrXrsby\n3w6TlHzE3yGpEriT9JsCyS7PU5zbThORbkAzY8w3hfb9FDgB7AP2AM8bYw6XP1ylVCD5afMB/vzp\nWvq1rs839w4gOjKMtxbt9HdYqgTuJH0pYps5/aJICPAS8EAR5XoB+UAToCXwgIi0OusEIuNEJFFE\nEtPS0twKXCnlXyt3H+buj1fRoXEtpt6cQP2akYzp3Zw567RuP5C5k/RTgGYuz+OAVJfn0UAn4GcR\n2QX0AWY7G3PHAN8ZY3KNMQeBJUBC4RMYY6YaYxKMMQmxsaVOEqeU8rMt+49z67QVNK5dnWm39qRm\npJ278Zb+8YSI8O6S3/wcoSqOO0l/BdBGRFqKSAQwGphd8KIx5qgxJsYYE2+MiQd+AYYbYxKxVToX\niBWF/ULY7PHfQinlMykZWdz87nKqhYfywW29iKkZefq1xrWrM/y8JsxYkczRrFw/RqmKU2rSN8bk\nAROBucAmYKYxZoOIPCkiw0vZfQpQE1iP/fKYZoxZW8GYlVJ+cigzm5vf+ZWTOfl8cHsvmtWrcVaZ\nOwa2Iisnn4+WuzXTr/Ixt+bTN8bMAeYU2vb3Ysqe7/I4E9ttUylVyWVm53HreyvYe+QkH9/Rm3aN\nahVZrkOTWgxsE8N7S3dxx8CWRIaF+jhSVRIdkauUKlV2Xj7jP0xkQ+ox/nNDdxLi65VY/s6BrUg7\nns1Xa1JLLKd8T5O+UqpE+Q7D/TOSWLL9EM+O7MKF7RuWus/ANjG0axTNWwt3YowptbzyHU36Sqli\nGWP4+1fr+d+6ffz18vaM7BHn1n4iwrhBrdh2MJOft2o37ECiSV8pVayXf9zGx8v3MH5wK+4cdNYQ\nmxJd0aUJjWpV462FOlgrkGjS96CcPIe/Q1DKYz5YtotX5m3jmh5xPDy0XZn3jwgL4db+8SzdcYj1\ne496PkBVLm713lFnyslzsCMtk837j7Fp33E27bM/j2Tl8LcrOjC2X7y/Q1SqQr5OSuWx2Ru4qH1D\n/nV1Z0SKGphfuut7N+e1n7bz1qKdvDK6m4ejVOWhSb8Uacezncn9GJv3HWfjvmPsSMskN982TkWE\nhtCmYU0GnxtL6pGTPPH1BuLqVnersUupQLRoWxr3z1xDQou6TB7TjbDQ8lcI1KoWzuiezZi2dBd/\nGdqOpnWqezBSVR6a9F3sTMtkTfIRNu///eo9PTP79OsNa0XSvnEtzm/bgPaNo2nfuBYtY6IId/5R\nZOXkcd2bv3DvJ6uZNaEvHZvU9tevolS5JCUfYfyHK2kdW5O3x/akWnjF+9jfOqAl05buYtri33j0\nirNmZVc+JoHWnSohIcEkJib6/Lyzk1KZ9MlqwNZFntuwJu0a1aJ941q0bxRNu8a1qBcVUepxDhw7\nxZVTlmAMfHlPfxrVrubt0D1iy/7jPPPdZi7v3JiR3ZuW+3ZeVV7bD2ZyzRtLiYoM4/O7+tGgluc+\nu/dNX82PGw+w9JELqV093GPHVb8TkZXGmLPmNjurnCZ92Jh6jKtfX0LnprX551WdaRkTVaFb2k37\njjHq9aXEx0Qxc3xfoiID94bKGMP0Fck8PnsD+Q5DnsNwQbsG/OvqzjT04B+9Cmw5eQ4uenEBWTl5\nfDqhH/ExUR49/vq9R7nitcU8clk7xg9u7dFje8Kp3HwSd2UQV7e6x393X3E36Qd9752MEzmM+zCR\nOtUj+M8NPWjTMLpCCR+gfeNaTB7TnU37jnHf9NXkOwLri7XA8VO5TJq+hkc+X0evlvVY+vAF/O2K\nDizdkc7FLy7gs5UpPhtYY4xh9Z4M9hzK0sE8fvDVmr3sOZzFc6PO80rS69S0Nv1a12fakl0B08st\n+XAWHy7bxa3TfuW8J77nxneWc8ELP3P/zDXsSj/h7/C8Jqiv9PPyHdwybQW//naYGeP70K15XY8e\n//2lu3hs9gZuH9CSvwVYXea6lKNM/GQVKRknuf/ic7lrcGtCQmyVzm/pJ/jzrCQSd2dwYbsG/NOL\nV/3GGBZsTePFH7ayNsV264uNjiShRV16tKhLQnw9OjapdbrdRHmew2EY+spCQkT49r6BXqvam7/l\nILdOW8GL157H1d3dG+TlSbn5DhJ3ZTB/y0Hmbz7ItoOZALSoX4MhbRswsE0Mv+w8xAfLdpPnMIzs\n3pR7L2hT5KRygUird9zwrzmbeHPhTp4d2YVrezYrfYdyeHz2Bt5buounruzETX1aeOUcZWGM4b2l\nu/jnnE3E1Izkteu7FTmPSr7DMG3Jbzw3dwuRYSE8PrwjV3XzbF3/0h3pvPD9VlbuzqBpnercPaQ1\nDgMrdx0mcXcGKRknAagWHkLXZnVIaFGPHvF16d68rtYLe9BPmw9w23uJXk/GxhgueWkhoSHe/XJx\ndfD4KX7eksbPWw6yaGs6x7PzCA8Veresz5B2DRjSNpZWsTXP3OfYKV5fsIOPl+/B4TBckxDHPUPO\nIa5uYCd/TfqlKGi4vbFPc/5xZWevnSffYRj3QSI/b03jnbEJnN+2gdfOVZojWTn8+dO1/LDxABe1\nb8Bzo86jbimN0zvSMvnLp2tZuTuDi9o35J9XdapwA1/irsO88P1Wlu08RKNa1Zh4wTlcm9CMiLAz\nr+YPHDtF4q4MVuw6zMrdGWzcd4x8h0EE2jaMdt4J1CWhRT3i6lbXxudyuu7NZSQfzmLBX4Z4/Y5q\nZmIyf/l0LR/e3ouBbTy/YJLDYUhKOcJ8Z6IvuHtsWCuSIW0bMKRdA/qfE3N60ZeS7D96itd/3s4n\nvyZjMFzXsxn3DDmHxrUDs9upJv0SuDbcfnxHn7OSjaedyM7jmjeWsedwFp/e1bfYKWm9aeXuw9z7\n39WkZWbz8GXtua1/vNtJ0vWqv1p4KE8M78iIrk3KnGSTko/w4g9bWbA1jZiakdx9fmvG9G7udrfA\nE9l5JCUfIXG3/SJYvecImdl5ADSIjqRHi7rUqRGBiF3jM0Tk9OOCWEWc252P5fRjIUSgVvVw6kdF\nEBMdSUxUJDHREdSLiqiy0wOvST7ClVOW8Oiw9twxsGzTLJRHdl4+A56ZT7tG0Xx4e2+PHfdkTj7P\nfLeZr5NSOXQihxCBbs3rckG7BpzfNpYOjWuV+6Ig9chJpszfzszEZATh+l7NuHvIOQHX0UGTfjEy\nTuQwfMpicvIcfH3vABpE++Y/bt/Rk1w5ZQmhInx5T3+PdocricNheGPhDl74fitN61Rn8phudImr\nU65j7UjL5M+zkli15wgXd2jI01d1cuv925h6jBd/2MqPmw5Qt0Y44we35ua+LagRUbFeTfkOw5b9\nx1m521YHrUk+wonsfMBgjF3I2WGcj42xCzsb5zZwlil43W7PK6bRPbpaGLE1I6lfM4L6zi8D+zOS\nmKgI6teMpEF0JC3q16hUdxx3f7ySRdvSWfbIhW5d/XrClPnbeW7uFr69byDtG1f8Amj3oRNM+GgV\nm/cf44ouTbiofQMGtYkt9S62rFIyspgyfzuzElMIDRFu6N2CCee38lkOKY0m/SLkOwy3TPuV5Tu9\n03BbmvV7j3LNG8to07AmM8b1pXqEd68e0zOz+dOMNSzals6wLo3519WdqVWtYnXh+Q7DO4t38vz3\nW6kRYa/6h59X9FX/tgPHefnHbfxv3T6iq4UxbmArbh3Q0mfJpayMMWTl5HMoM4e0zGwOZWZz6EQO\n6cedPzOzSc/M5lBmDodO5JCRlUPhP58RXZvw4rVdCQ0J/MS/K/0EQ174mQmDW/NQOebWKa+jWbn0\n/fc8hnZqxIvXdq3QsX7afIA/Tl+DiPDy6K4M8UH1afLhLF77aRufrdpLeKhwU58WjB/c+oxlI/1B\nk34R/vXtJt5csJNnRnbmup7NvXKO0vyw8QDjPkzk0g6N+M8N3U/3mPG0pdvTuW/GGo6dzOXvf+jA\nmF7NPXoFuv1gJn/+NInVe45waceG/OPKzsRG2w/9b+kneOXHrXyVlEqN8FBuG9CSOwa0onaNqtX4\nmpfv4HBWDocy7RfCku2HeGPBDq7u3pTnR53ntf9bT3n0y3XMXJHC4oeG+OzOs8ATX2/gw2W7WfTQ\nkHLVkec7DK/M28ar87bRoXEt3rixB83r+7ahdVf6CV79aRtfrt5LZFgoN/drwZVdm9KsXg2/XNho\n0i/k66RU7v1kNTf0bs7TV3mv4dYd7yz+jae+2cj4Qa145PL2Hj12wR/Daz9to1VMFJPHdPfILXRx\n53p70U5e+GErURGh/GVoO1bvyTh9BTS2XzzjB7V2ayRzVfHKj9t46cetjO7ZjH9e1TlgE/+hzGz6\n/fsnruzalGdGdfH5+ZMPZzH4ufncOagVj1xWtr+BI1k53Dd9DQu2pjGyexxPX9XJI9NFlNeOtExe\nm7eNr5JST9/51akRTrO6NYirW935r8YZP70xYNPdpB+Y99ketmnfMf7y6VoSWtTlsT909Hc43NY/\nnl3pJ3hz4U7iY6K4vpdn7jr2Hz3FfdNXs/y3w4zsHseTIzp6dTRwaIgwfnBrLmzfgAdnreWRz9cR\nERbCzX1bcNf5rQOmrtOXJl14Drn5DibP3054aAhPjugYkHX87y/bTXaeo8xz5HtKs3o1uLxzY/77\nyx4mDjmHaDerHdfvPcqEj1Zy4Ngpnr6qk8fvYMujdWxNXh7djT9dfC7r9x4jOSOLlIwsUjJOsu1g\nJvO3HORU7pkD0upFRRTxhVCdc2KjvX7H4lZGEJGhwCtAKPC2MebfxZQbBcwCehpjEp3bugBvArUA\nh/O1Ux6I3S1HsuyI21rVw/jPjd293lPHHSLCY3/owJ7DWTz65Xqa1a3BgDYxZT5OVk4eSclHWbUn\ng9V7Mli+8zD5xvDCNee5vcKRJ5zTIJpPJ/Rl0bZ02jWODtgubb4gIjxwybnk5DuYunAnEWEhPDqs\nvd8Tk6usnDw+XLaLi9o35JwGNUst7y3jBrXim7X7mLEi2a2eQ7MSk3n0y/XUi4pg5vi+Pm+TK02L\n+lG0qH/2aGZjDOmZOae/CFIyTjq/GE6yef9xftx08PQo5WGdGzPlhu5ejbPUpC8iocAU4GIgBVgh\nIrONMRsLlYsGJgHLXbaFAR8BNxljkkSkPpDrwfhLlO8w3PvJag4czWb6+D4BdeUZFhrC5DHduOaN\nZdz18Uo+v6sfbRpGF1veGENKxklW7s5g1R77b9O+46eneGgVG8XQTo0YP7i1X/6Qw0JDGNLOf2MQ\nAomI8Mhl7cjJc/DO4t8IDw3hoaFtAybxz0pMISMrlwmD/XOVX6BLXB16t6zHu4t/Y2y/+GLHCGTn\n5fPE1xv57/I99G1Vn9fGdPN7o2lZiAix0ZHERkcW+UXlcBjST2STfPgk1cK9f1HqzpV+L2C7MWYn\ngIhMB0YAGwuVewp4FnjQZdslwFpjTBKAMeZQhSMug+fmbmHRtnT+fXVnugfYVQFAdLVw3rmlJ1dO\nWcKt763gi7v7n24MPZWbz7q9R22S353Bqj1HTk/zXCMilK7N6nDX4NZ0b1GHbs3qerx7mqqYgru5\nnHwHbyzYQURYCPdffK6/wyIv38Hbi3fSvXmdIkdi+9q4Qa24/f1E5qzbx4iuTc96PfXISe76eBVJ\nyUeYMLg1D15yboXnxgo0ISFCg+hqPrsodSfpNwWSXZ6nAGeMqhCRbkAzY8w3IuKa9M8FjIjMBWKB\n6caYZysYs1u+WZvKGwt2MKZ3c0Z7qM7cG5rWqc47YxO49s1l3PH+Crq3qMuq3RlsSD12us94fP0a\nDGoTQ7cWdenevA5tPTApnPI+EeEfIzqRm+fg1XnbiAwL4Z4h5/g1pm/X7yf58EkeHRYYc0ENaduA\n1rFRTF2486yuv0u3pzPxk9Xk5Dl448buDO3U2I+RVh3uJP2i7klPd/kRkRDgJeCWYo4/AOgJZAHz\nnC3M8844gcg4YBxA8+YVT9Cb9x/jz7PW0qNFXR4PgIbb0nSJq8PL13Xj7o9XsuXAcc6Lq8Odg1rR\nvXldujWvU6luZdWZQkKEf4/sQm6+g+fmbiEiNMRvjafGGKYu3EmrmCguDpCV3UJChDsHtuLhz9ex\nbMch+p0TgzHGzon13WZaxdbkjRt7+LXtoapxJ+mnAK6zkcUBqS7Po4FOwM/Ob+lGwGwRGe7cd4Ex\nJh1AROYA3YEzkr4xZiowFWyXzXL9Jk5HsnIY98FKoquF8foNgdFw646hnRqx/P8uok6NcJ1RsooJ\nDRGev+Y8cvMNT8/ZRHiocEv/lj6PY9mOQ6zbe5R/XR1YXUmv7NaU57/fwtRFO+kcV5sHZyUxd8MB\nhnVpzLMjuwT0ehSVkTvv5gqgjYi0BPYCo4ExBS8aY44Cp7ueiMjPwIPGmEQR2QH8RURqADnAYOxd\ngVcUNNzuO3qS6eP6+nzASUUV1OerqicsNISXR3clN9/B419vJDwshBt6+3bW1TcX7iSmZiRXdTu7\n7tyfqoWHMrZvPC/8sJXLX11E6pFTPDqsPbcPaBkwjd9VSamXlMaYPGAiMBfYBMw0xmwQkSedV/Ml\n7ZsBvIj94lgDrDLG/K/iYRft+e9tw+2TIzrRo0XgNdyq4BYeGsJrY7oxpG0sf/1iPTMTk0vfyUM2\n7TvGgq1p3No/3q8DmYpzY58W1IgI5WROPh/f0Zs7BrbShO8lVWZE7tLt6Yx5eznX92rOv67274hb\npUpyKjefOz9IZPH2dF66titX+uDK+/4Za/huw36WPXxhwE6HsSMtkzrVw6mvbVjlEnTLJfZqWY/H\n/9CBx4cHRq8EpYpTLTyUqTcl0Kdlfe6fuYb/rd3n1fOlHjnJ7KRURvdsHrAJH+zIVk343ldlkn5Y\naAi39G9ZZec9V1VL9YhQ3h6bQPfmdblv+mrmbtjvtXO9u/g3DHDbgHivnUNVHlUm6StV2URFhjHt\n1p50alqbif9dxU+bD3j8HEdP5vLJr3v4Q5fGAb/cn/INTfpK+VF0tXDev60XbRtFM+GjVXy33rNV\nPR8v382JnHzGDWrt0eOqykuTvlJ+Vrt6OB/d3pv2zsT/6JfrOJWbX+HjZuflM23JLga2iaFDE98v\n0akCkyZ9pQJAnRoRzJrQjzsHtuSjX/YwfPJituw/XqFjfrl6L2nHsxmvV/nKhSZ9pQJERFgIfx3W\ngfdv68XhE7kMn7yYD5ftojzdqh0OO+VCxya16H9Ofc8HqyotTfpKBZjB58by3R8H0qdVff721QbG\nfbiSjBM5ZTrGvM0H2ZF2gnGDdJCTOpMmfaUCUEzNSKbd0pNHh7Xn5y0HGfrKQpbuSHd7/6kLd9C0\nTnWGddaZKdWZNOkrFaBCQoQ7Brbii7v7ExURxg1vL+e5uZvJzXeUuN/K3Rms2JXBHQNb6hTc6iz6\niVAqwHVqWpuv7x3ANT3imDJ/B9e+uYzkw1nFlp+6cAe1q4dzbUKzYsuo4KVJX6lKICoyjGdHncfk\nMd3YfjCTy19ZxFdr9p5VbmdaJt9vPMDNfVvolMSqSJr0lapErujShDmTBtKmYU3um76GB2clcSI7\n7/Trby2ya/Le3Dfef0GqgKZJX6lKplm9Gswc35dJF5zD56tSuOK1xaxLOUra8Ww+W5XCqB5xujaD\nKpbe/ylVCYWFhnD/JW3pd04Mf5qxhqtfX0KXuDrk5ju4c6B/lmNUlYNe6StVifVpVZ9v7xvIBe0a\nsHJ3Bpd0aEjLmCh/h6UCmF7pK1XJ1akRwRs39uDnrWl0aVrb3+GoAKdJX6kqQEQY0raBv8NQlYBW\n7yilVBDRpK+UUkHEraQvIkNFZIuIbBeRh0soN0pEjIgkFNreXEQyReTBigaslFKq/EpN+iISCkwB\nLgM6ANeLyFmrj4tINDAJWF7EYV4Cvq1YqEoppSrKnSv9XsB2Y8xOY0wOMB0YUUS5p4BngVOuG0Xk\nSmAnsKGCsSqllKogd5J+UyDZ5XmKc9tpItINaGaM+abQ9ijgIeCJCsaplFLKA9xJ+kWtwHB6KR8R\nCcFW3zxQRLkngJeMMZklnkBknIgkikhiWlqaGyEppZQqD3f66acArnO0xgGpLs+jgU7Az84VehoB\ns0VkONAbGCUizwJ1AIeInDLGTHY9gTFmKjAVICEhoexrwymllHKLO0l/BdBGRFoCe4HRwJiCF40x\nR4GYguci8jPwoDEmERjosv1xILNwwldKKeU7pVbvGGPygInAXGATMNMYs0FEnnRezSullKokxJjA\nqk1JSEgwiYmJ/g5DKaUqFRFZaYxJKK2cjshVSqkgoklfKaWCiCZ9pZQKIpr0lVIqiGjSV0qpIKJJ\nXymlgogmfaWUCiKa9JVSKoho0ldKqSCiSV8ppYKIJn2llAoimvSVUiqIaNJXSqkgoklfKaWCiCZ9\npZQKIpr0lVIqiGjSV0qpIKJJXymlgogmfaWUCiKa9JVSKoho0ldKqSDiVtIXkaEiskVEtovIwyWU\nGyUiRkQSnM8vFpGVIrLO+fMCTwWulFKq7MJKKyAiocAU4GIgBVghIrONMRsLlYsGJgHLXTanA38w\nxqSKSCdgLtDUU8ErpZQqG3eu9HsB240xO40xOcB0YEQR5Z4CngVOFWwwxqw2xqQ6n24AqolIZAVj\nVkopVU7uJP2mQLLL8xQKXa2LSDegmTHmmxKOMxJYbYzJLvyCiIwTkUQRSUxLS3MjJKWUUuXhTtKX\nIraZ0y+KhAAvAQ8UewCRjsAzwPiiXjfGTDXGJBhjEmJjY90ISSmlVHm4k/RTgGYuz+OAVJfn0UAn\n4GcR2QX0AWa7NObGAV8ANxtjdngiaKWUUuXjTtJfAbQRkZYiEgGMBmYXvGiMOWqMiTHGxBtj4oFf\ngOHGmEQRqQP8D3jEGLPEC/ErpZQqg1KTvjEmD5iI7XmzCZhpjNkgIk+KyPBSdp8InAP8TUTWOP81\nqHDUSimlykWMMaWX8qGEhASTmJjo7zCUUqpSEZGVxpiE0srpiFyllAoimvSVUiqIaNJXSqkgoklf\nKaWCiCZ9pZQKIpr0lVIqiGjSV0qpIKJJXymlgogmfaWUCiKa9JVSKoho0ldKqSCiSV8ppYKIJn2l\nlAoimvSVUiqIaNJXSqkgoklfKaWCiCZ9pZQKIpr0lVIqiGjSV0qpIKJJXymlgohbSV9EhorIFhHZ\nLiIPl1BulIgYEUlw2faIc78tInKpJ4JWSilVPmGlFRCRUGAKcDGQAqwQkdnGmI2FykUDk4DlLts6\nAKOBjkAT4EcROdcYk++5X0EppZS73LnS7wVsN8bsNMbkANOBEUWUewp4Fjjlsm0EMN0Yk22M+Q3Y\n7jyeUkopP3An6TcFkl2epzi3nSYi3YBmxphvyrqvc/9xIpIoIolpaWluBa6UUqrs3En6UsQ2c/pF\nkRDgJeCBsu57eoMxU40xCcaYhNjYWDdCUkopVR6l1uljr86buTyPA1JdnkcDnYCfRQSgETBbRIa7\nsa9SSikfcudKfwXQRkRaikgEtmF2dsGLxpijxpgYY0y8MSYe+AUYboxJdJYbLSKRItISaAP86vHf\nQimllFtKvdI3xuSJyERgLhAKvGuM2SAiTwKJxpjZJey7QURmAhuBPOAe7bmjlFL+I8acVcXuVwkJ\nCSYxMdHfYSilVKUiIiuNMQmlldMRuUopFUQ06SulVBDRpK+UUkFEk75SSgURTfpKKRVENOkrpVQQ\n0aSvlPKMDV9C+jZ/R6FKoUlfKVVxh3bArFvgmz/5OxJVCk36SqmK++U/gIFdi2D/en9Ho0qgSV8p\nVTFZh2H1x9DuCgirDr++6e+IVAk06SulKibxHcg7CUP+Cl2uhbUz4cQhf0eliqFJXylVfnnZ8Otb\n0PpCaNgBek+AvFOw6n1/R6aKoUlfKVV+62ZB5gHoN9E+b9gBWg6GFW9Dfq5/Y1NF0qSvlCofY2DZ\nFGjYCVoN+X177wlwbC9s+tp/saliadJXSpXPjp/g4Eboew+Iy8qo514KdeNhuTboBiJN+kqp8lk2\nGWo2hE4jz9weEgq9xkHyL5C62j+xqWJp0ldKld2BDfZKv9c4CIs8+/VuN0JETfjlDd/HpkqkSV8p\nVXbLpkB4DUi4rejXq9WGrmNg/Wdw/IBvY1Ml0qSvlCqb4/ttX/yuN0CNesWX6zUOHLmwcprvYlOl\n0qSvlCqbX6eCIw/63FVyuZg2cM7FsOIdyMvxTWyqVG4lfREZKiJbRGS7iDxcxOsTRGSdiKwRkcUi\n0sG5PVxE3ne+tklEHvH0L6CU8qGcE5D4LrQbBvVbl16+zwQ4cRA2fOH92JRbSk36IhIKTAEuAzoA\n1xckdRf/NcZ0NsZ0BZ4FXnRuvwaINMZ0BnoA40Uk3kOxK6V8bc1/4WQG9J3oXvlWF0D9NrD8dduv\nX/mdO1f6vYDtxpidxpgcYDowwrWAMeaYy9MooOB/1wBRIhIGVAdyANeySqnKwpFvZ9Ns2gOa93Fv\nn5AQ6D3edt1MWeHd+JRb3En5DXmSAAAZpElEQVT6TYFkl+cpzm1nEJF7RGQH9kp/knPzp8AJYB+w\nB3jeGHO4iH3HiUiiiCSmpaWV8VdQSvnElm/h8E57le86GKs0510PkbXhl9e9F1tFOfLhszvtPEJV\nnDtJv6j/3bPu04wxU4wxrYGHgEedm3sB+UAToCXwgIi0KmLfqcaYBGNMQmxsrNvBK6V8aNlkqN0c\n2g8v236RNaH7TbDxKzi61zuxVdSyKbBuJsz9Kxz+zd/ReJU7ST8FaObyPA5ILaH8dOBK5+MxwHfG\nmFxjzEFgCZBQnkBVFZd7Er65X+drCVQpK2HPMttjJzSs7Pv3uhOMw07DHGjStsJP/7ATxYWEwfeP\nlr5PJeZO0l8BtBGRliISAYwGZrsWEJE2Lk+HAQULZe4BLhArCugDbK542KrK+f5RmxBm3AizboUT\n6f6OSLlaNtlW0XS/qXz7142HtpdD4jT7BR8oHPnw1d0QUQOufgsG3g+bv4GdP/s7Mq8pNekbY/KA\nicBcYBMw0xizQUSeFJGC+7yJIrJBRNYA9wNjndunADWB9dgvj2nGmLWe/iVUJbd5jp2Kt/ddMORR\ne7U/pTes/1x7fASCI3ts1UyPsRAZXf7j9JkAJw/Duk89F1tFLZtiG5gvfx6iG9r2irrx8N0jkJ/n\n7+i8QkyA/VElJCSYxMREf4ehfOXYPni9H9SOgzt+tPO4HNhor75SV0P7P8CwF6FmA39HGry++z+7\nBOJ9Sfb/qbyMgdf720bgCYvL1hjsDWlb4Y0B0OZiuO6j3+PZ9A3MuAEuew56j/NvjGUgIiuNMaVW\nn+uIXOU/Dgd8Md6utDTq3d8n7mrYAW7/ES56HLZ+D1N62WH/AXaBEhROHYVVH0DHqyqW8MEm1d7j\n4cB62LXYM/GVl2u1zrAXz/wCajfM1u/Pf9qu/1vFaNJX/rP0VfhtAQz9tx2y7yo0DAb8CSYsgvrn\nwOd3wvQxdt4X5Tsr34ec4+4PxipNl2uhel1Y7ufZNwtX67gSgcuegezjNvFXMZr0lX/sXQU/PQUd\nRkD3m4svF9sWbpsLlzxtp/Kd0suOCtWrfu/Lz7XJOX4gNOnqmWOGV4cet8CWOZCx2zPHLKuC3jrt\nrjh7LYACDdpDzzvslBP71/s2Pi/TpK98LzsTPrsdajaCP7xSet1uSKhdg3XCEmjQAb68Cz6+JnD7\nfFcVG7+yyx566iq/QM87AIEVfhgIVVK1TmHnPwzV6sB3D1epiwxN+sr3vn3IDoC5eqq91XdXzDlw\nyxwY+gzsXgL/6WOrH6rQH2TAMAaWvmbnzWlziWePXTsOOgy3bQU5Jzx77NKUVK1TWI16cMFfYdci\n2DS75LKViCZ95VvrP4M1H8GgByG+f9n3DwmxXf/uWgKNz4OvJ8GHV9luhcpzdi+BfWug7932Pfe0\n3hNsI3HSJ54/dnHcqdYprMetduH3uY8G1viCCtCkr3znyB74+k8Q1xMGP1SxY9VrBTfPhmEv2Cu3\n//S1ff0dDs/EGuyWToYa9e28Od7QrDc07moXT/fFnVpZqnVchYTajgZH99g7nypAk74naTVD8fLz\n7IRWxmFHPoaGV/yYISG2fviupRCXAP97AH74W8WPG+zSt8HWb+17G17dO+cQsVM6pG+1DfTetmyy\n+9U6hbUcaDscLHoRjqZ4Jz4f0qTvCXtXwefj4J9NbH9ydbZFz0PyL3DFi1CvpWePXbcF3PSlTVLL\nJtuRvKr8lk2B0Ejoead3z9PxKohqYK/2vSltK/z0dNmqdQq7+CnAwA+PeTS0M+Se9EkVkib98srP\ns8nlnUvgrSGw+X8Q3Rhm3wv7dKaJM+z5BRY8A11G237a3iACl/4L4nrBVxPhoE7xVC4n0m09+3nX\nQU0vz3gbFmkXVt82Fw7t8M45XKt1rnip/KOA67aAfpNg/aewe5lnYwTI2AXvXGwnHfQyTfpllXXY\n3ua90gU+vRUyD9hkc/9GuO07qF7PThpWBUfylcvJI7Zap05zuPw5754rLAKufd/+gc+4EU7pej1l\nlviuHSHt6W6axUm4DULCvXe171qtU9GpPAb8EWo1hW//Yr9MPGX7PJh6vm3z6niV545bDE367jq4\nCb6+D17sAPOesOuDjv4E7l1lezhUq20/VNd+AMdSbXVPsDcqGgPf/Mn29R75DlSr5f1z1moCo6bZ\nxT6+ukfbWcoi95Rd9LzNJXZQnC9EN4ROV8Oajz3/Je2Jah1XEVFw8ZOwfy2s/qjixzMGFr0AH42E\n6CZw53w418PdY4ugSb8kDgds+Q4+GGH7hCdNhy7X2IbDsV9Du8tt676rZj3hsn/D9h9gwb/9E3eg\nSPoENnwOQ/7PNrT6SsuBcNFjtm/1ssm+O29lt24mnEiDvvf49ry9x0NOpk38nuKpap3COo2E5n1h\n3pP2Lra8Th2zd6PznrRfenf84N5C8x6gSb8o2cfhlzdgcg/45Dp7xXDh3+FPG2H4a9CwY8n7J9wO\n542x9dhbvvNNzIHm0A7434PQYoCdQ8fX+k2yM3T+8Jj/J/eqDBwO24DbsLOdbMyXmvawbTHL3/Tc\n3bEnq3VcidgunFmHYGE5qyvTtsLbF9rlJy/9p70LjojyXIyl0KTv6vBvdh7tFzvAdw/Zfsoj34E/\nroWBD0BUffeOI2J7qTTqYqt5vNVIFajycuw0C6HhcPWbZ98N+YIIjPiP7c8/6xZb5aaKt24WpG2G\ngX/yz5THfSZAxm924fXMgxU7VtoWz1brFNakq11MZvkbNoGXxaZv4K0LbJvfzV/Zuyofv986n36B\npBl2ThcR25jS+y6I61GxY2bsgjcH22Hnt/9gbzWDwQ+PwZKX4doP7XB7fzq4Cd66EBp1grHf2MZe\ndaa8bJicYOeZGbfAOyNwS5OfC28MhLRN9nndeHv136yXHczXsJN7yzQ68m2PusM74Z7l3luHITMN\nXuthq3Nv+LT0xO3ItzN2LnrB3tlc+yHUburRkNydT78ci11WQas+tF0t4wfYgUO1GnvmuHXjYdQ7\n8NEo2wh89VT/LxzhbTt/hiWv2JkU/Z3wwc6WOOI1+PQ2O3Drsmf8HVHgSZxme47c+LJ/Ej7Yu8Lx\nCyB1DaT8Csm/wm8LbTsDQHgNZzVQTzuaN65n0XfeyybD3kR7h+7NhXdqxsL5D8Hc/4Nt38O5lxZf\nNuswfHYH7JgH3cfaXmwFa0f4gV7pr3jbjuRsfSGM/tg7IxAXPAfz/wGXPWsbraqqE4fgjf52Sb1x\nP/u0nrJU3z1iqw5GvgOdR/k7msCRfRxe6WoXrrl5dmBdlBgDR5PtF0Dyr/bLYP86cDiXMazX2n4B\nNOtp7wpCQu2ddeGVsLwlL8eu+mYccPcvRd9F7ltrG2yP77PJvsctXgtHr/Td8cvrdtrUc4fCNe9D\neDXvnGfgA7B3pb0qaNQFWvT1znn8xeGAgxvgxydsA9eYGYGV8MF2tUtdbe/oGnSwSU7ZOXay0uHC\nxwMr4YONp05z+6/gizony04El7wcklfYXnJJ//19n+r1PNtbpyRhEbZR9+ORsPx16H/fma8nzbB3\n+DXqwa3fVby62EOC90p/8cvw42O2h8fId71f13vyiB25m3MCxi+E6EbePZ83OfJtX+VdS+xsjLuX\nwiln97VAXlf0+H54cxBE1IRx8+3YimCWmQavdoXWF8B1H/o7mvIxxjYAJ6+wF1ZtL4PWQ3wbw3+v\ns38L96604w7yc+H7R21Db4sBcM173h/djPtX+m4lfREZCrwChAJvG2P+Xej1CcA9QD6QCYwzxmx0\nvtYFeBOoBTiAnsaYU8WdyydJf8GztlGl00i46k3PTP7ljgMb4O2L7JTAY7/23XkrKj8X9iXZro+7\nl9hpFbKdA2nqtYIW/W17SIv+UKeZf2Mtze6l8N4VNjn4ogogkH37EPz6lm3wLLxcpXLfoR0wpTd0\nuc527Z51C+xZakc1X/SEew3QHuCxpC8iocBW4GIgBVgBXF+Q1J1lahljjjkfDwfuNsYMFZEwYBVw\nkzEmSUTqA0eMMcWOYfZq0jfGJvuFz9l5YK78j++7E6771HZn7H2XHcQViPJyIHWVS5JfDrnOxS5i\nznVJ8v3sCNjKZtkUW9V20eP+GUPgKifLVjul/Gq/WHvcAq3O9/55M3bBawnQdQwMf9X756vqvv+b\nXfM5qoEdaDb8NZ+3HXmyTr8XsN0Ys9N54OnACOB00i9I+E5RQME3ySXAWmNMkrPcIffC9wJj4Ie/\n2/+YbjfZZfr80X+88yhISbR1gHEJgdGomHvK3hrvXmITffKvkOec7a9BB5sY4vvbZO/NHhG+0udu\nO3Bn3pPQpDu08tFgJGPgyG5bFVHQQ+XA+t8bJsNrwM4FdoEYb3+Zzv+n/fyf/7B3zxMsBv3ZjnUI\nqwY3fWG7CAcod5J+UyDZ5XkK0LtwIRG5B7gfiAAucG4+FzAiMheIBaYbY56tUMTlYYztvbH8dTta\n9vLn/dc1DeCSp+xV3ex7bZfC0kb4elruSZtwdi+xdZEpKyA/GxD7Ye1xi03yzfu5PyCtMhGB4ZPh\nwEbblXP8Qo/3mQbs+5y62tnzZIX9ecI58Cg8Cpp2t41/cc6+6CcP294nn90JY2d776Jk/3o7BXj/\nSZXzTi0QVatlp2cJr+69NQg8xJ2kX1Sl51l1QsaYKcAUERkDPAqMdR5/ANATyALmOW9B5p1xApFx\nwDiA5s2bl+kXKJXDAXMesLMH9rnbDnv2dz1uaLht3HlzEEy/wXZvrF7He+fLOWF7OxQ0vKYkgiMX\nJMT2Jup1p62uad6nbGvWVmaRNW2d/ltDYObNcOucivWddjjsVfzelcV0L2xlG0wLuhc26HB2XW9U\nfRj2vB0kuPB52w/cG+Y9aZOUv6u2qpoa9fwdgVvcSfopgGvrXBxQ0pj26cDrLvsuMMakA4jIHKA7\ncEbSN8ZMBaaCrdN3K3J3OPLtGqqrP4L+f7R1uP5O+AWiG9oZOd+7HL4Yb2fs9NTdR/ZxWw+/a5FN\n8qmrbfKRUDuEvO/dtldB897B3YMl9lwYMQVmjbV1/MNeKH2fvBw72jN9ix2Cn7bZPk7f/nuVWMFA\non6Tfh9RGhXjXkznXQ875tvJ+loOtO0mnrR7qZ2//qLHg+cLXp3BnaS/AmgjIi2BvcBoYIxrARFp\nY4zZ5nw6DCh4PBf4i4jUAHKAwcBLngi8VPl5dpa9tTNg8MO27jJQEn6B5r3tXPzf/tkOzx7857Lt\nf+ooHEm2A1iOJNuua3t+sVVHJh9Cwmyddb9JtrqmWW87cEr9ruOVsPdeu/5pXE84b7TdnnPCLuWX\nttWZ4J3/Mn77/eodoHZz++URP8j+bNINGnQsf4+Ngnmb9ibaUZwTFnvuCtI4V36Kbgy9qvAgQVWi\nUj+Zxpg8EZmITeChwLvGmA0i8iSQaIyZDUwUkYuAXCADW7WDMSZDRF7EfnEYYI4x5n9e+l1+l59r\n/2A2fgkX/A0GPej1U5ZbrzvtH/j8p23CaHOR3e5w2GlujybbIfIFif1oyu+Ps4+eeazQSFtPPPB+\n2+jarFfgDZIKRBc+DntXw9d/dE48ttUuhF1AQm31TGxbO7VETFub4GPO9c77GxkNo96Fty+2q4CN\n/tgzFyxb5thqpz+8EjzzQKmzVL3BWXnZtnFu8zdwyT+g372eC85bcrLsUmlHk6FxV2diT3E2rrqI\nrG37wdduZidxK3hcp7n9GRXr3wbqyizzoG1fyTvpTOptbVKPbWcTvj8maivoWnr58/bioCIc+XbK\nAEce3L3cZ33Hle8E5zQMuadg5k12AqTKNM9NRA07IvLzcbZaoXEXaDfMmczjnIm9WXDXv3tbzQZ2\nIYtA0uduO4Hd3L/aRvZGnct/rKRPbPvDNe9rwg9yVedKPycLpo+BnfPhipch4VbPB6eUr51Ih9ed\nk9iNX1C+6qTcU3Ya4JoN4M6fAq9tS3mEu1f6VacuYF8S7Flme2NowldVRVSMnZL70Ha7IHd5rHgb\njqUEVu815TdVJ+m36AuT1kC3G/0diVKe1Wqwnal19Ud2Go+yOHUUFj1vxwj4auSxCmhVJ+mD5xY/\nUSrQnP+I7XL79R/tOAF3LXkVTmbAhY95LzZVqVStpK9UVRUaBiPftr2zPr3dDhIrzfH9duGYjlfb\nQXlKoUlfqcqjTnM7Z1DqKvjpqdLLL3gW8nPggke9H5uqNDTpK1WZdBhuJw1c+ips+7H4cod2wKr3\n7Zqs9Vv7Lj4V8DTpK1XZXPq0nerhywlw/EDRZeY/DaERMNhLk7apSkuTvlKVTXh1O01DdiZ8Mc5O\n2eEqdQ2s/8wO7opu6J8YVcDSpK9UZdSgHVz2jB2xu+TlM1+b94SdQbP/JL+EpgKbJn2lKqvuN0PH\nq+Cnf9jVuMCuvLXjJ9uvX6ftUEXQpK9UZSViZ8ys3RQ+u832x//xcagVBz0rOEGbqrI06StVmVWr\nDSPfhWOp8M6ltjvnkEcgvJq/I1MBSpO+UpVds562L376FjsV9HnX+zsiFcB0jlWlqoJ+99nV4tpc\n5L0F1VWVoElfqaogJKTsy22qoKTVO0opFUQ06SulVBDRpK+UUkHEraQvIkNFZIuIbBeRh4t4fYKI\nrBORNSKyWEQ6FHq9uYhkisiDngpcKaVU2ZWa9EUkFJgCXAZ0AK4vnNSB/xpjOhtjugLPAi8Wev0l\n4FsPxKuUUqoC3LnS7wVsN8bsNMbkANOBEa4FjDHHXJ5GAadXWxeRK4GdwIaKh6uUUqoi3En6TYFk\nl+cpzm1nEJF7RGQH9kp/knNbFPAQ8ETFQ1VKKVVR7iR9KWKbOWuDMVOMMa2xSb5gqZ4ngJeMMZkl\nnkBknIgkikhiWlqaGyEppZQqD3cGZ6UAzVyexwGpJZSfDrzufNwbGCUizwJ1AIeInDLGTHbdwRgz\nFZgKICJpIrLbzfiLEgOkV2B/b9P4KkbjqxiNr2ICOb4W7hRyJ+mvANqISEtgLzAaGONaQETaGGO2\nOZ8OA7YBGGMGupR5HMgsnPALM8bEuhN4cUQk0RiTUJFjeJPGVzEaX8VofBUT6PG5o9Skb4zJE5GJ\nwFwgFHjXGLNBRJ4EEo0xs4GJInIRkAtkAGO9GbRSSqnycWvuHWPMHGBOoW1/d3l8nxvHeLyswSml\nlPKsqjgid6q/AyiFxlcxGl/FaHwVE+jxlUqMOasjjlJKqSqqKl7pK6WUKkalTPpuzAUUKSIznK8v\nF5F4H8bWTETmi8gmEdkgIme1d4jI+SJy1DlX0RoR+XtRx/JynLtc5ktKLOJ1EZFXne/hWhHp7qO4\n2rq8L2tE5JiI/LFQGZ+/fyLyrogcFJH1LtvqicgPIrLN+bNuMfuOdZbZJiJe6eRQTHzPichm5//f\nFyJSp5h9S/wseDG+x0Vkr8v/4+XF7Fvi37sX45vhEtsuEVlTzL5ef/88yhhTqf5hexDtAFoBEUAS\n0KFQmbuBN5yPRwMzfBhfY6C783E0sLWI+M4HvvHz+7gLiCnh9cux8yUJ0AdY7qf/6/1AC3+/f8Ag\noDuw3mXbs8DDzscPA88UsV897DQk9YC6zsd1fRTfJUCY8/EzRcXnzmfBi/E9DjzoxmegxL93b8VX\n6PUXgL/76/3z5L/KeKVf6lxAzufvOx9/ClwoIkWNLPY4Y8w+Y8wq5+PjwCaKmLaiEhgBfGCsX4A6\nItLYxzFcCOwwxlRksJ5HGGMWAocLbXb9nL0PXFnErpcCPxhjDhtjMoAfgKG+iM8Y870xJs/59Bfs\nwEq/KOb9c4c7f+8VVlJ8ztxxLfCJp8/rD5Ux6bszF9DpMs4P/VGgvk+ic+GsVuoGLC/i5b4ikiQi\n34pIR58GZhngexFZKSLjinjdrTmXvGw0xf+h+fv9A2hojNkH9sseaFBEmUB4HwFuo/iZbkv7LHjT\nRGf107vFVI8Fwvs3EDhgfh+AWpg/378yq4xJ3525gNyaL8ibRKQm8BnwR3PmLKQAq7BVFucBrwFf\n+jI2p/7GmO7YKbPvEZFBhV7363soIhHAcGBWES8HwvvnrkD4LP4VyAM+LqZIaZ8Fb3kdaA10BfZh\nq1AK8/v7B1xPyVf5/nr/yqUyJn135gI6XUZEwoDalO/WslxEJByb8D82xnxe+HVjzDHjnITO2IFv\n4SIS46v4nOdNdf48CHyBvY12VdY5lzztMmCVMeZA4RcC4f1zOlBQ5eX8ebCIMn59H50Nx1cANxhn\nBXRhbnwWvMIYc8AYk2+McQBvFXNef79/YcDVwIziyvjr/Suvypj0T88F5LwaHA3MLlRmNr9PBTEK\n+Km4D7ynOev/3gE2GWMKLyZTUKZRQRuDiPTC/j8c8kV8znNGiUh0wWNsg9/6QsVmAzc7e/H0AY4W\nVGX4SLFXV/5+/1y4fs7GAl8VUWYucImI1HVWX1zi3OZ1IjIUO+vtcGNMVjFl3PkseCs+1zaiq4o5\nrzt/7950EbDZGJNS1Iv+fP/Kzd8tyeX5h+1ZshXbqv9X57YnsR9ugGrYaoHtwK9AKx/GNgB7+7kW\nWOP8dzkwAZjgLDMRu6hMEraBrZ+P379WznMnOeMoeA9dYxTsimk7gHVAgg/jq4FN4rVdtvn1/cN+\nAe3Dzi+VAtyObSeah51gcB5Qz1k2AXjbZd/bnJ/F7cCtPoxvO7Y+vOBzWNCjrQkwp6TPgo/i+9D5\n2VqLTeSNC8fnfH7W37sv4nNuf6/gc+dS1ufvnyf/6YhcpZQKIpWxekcppVQ5adJXSqkgoklfKaWC\niCZ9pZQKIpr0lVIqiGjSV0qpIKJJXymlgogmfaWUCiL/D8Af4ZE6hcQnAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x20c3d5e1080>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(train_scores)\n",
    "plt.plot(valid_scores)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Neural networks are stochastic! As we can see, the validation score fluctuates quite drastically. This also has to do with our small validation set size. \n",
    "\n",
    "One way to combat such fluctuations is by creating an ensemble."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "ensemble_preds = [nn.predict(X_valid) for nn in nns]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(20, 199, 33)"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ensemble_preds = np.array(ensemble_preds); ensemble_preds.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.46993920498694269, 0.16444827050387467)"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r2_score(y_valid, ensemble_preds.mean(0)), rmse(y_valid, ensemble_preds.mean(0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.3647901085803823, 0.17688533705390513)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(valid_scores), np.mean(valid_rmses)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Again, we see something quite weird. The R2 score for the ensemble mean predictions went up drastically, but the RMSE only by a little bit.\n",
    "\n",
    "Another thing to pay attention to when doing ML: the metric. What do you actually want from the dataset?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Now with Keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "nn = Sequential([\n",
    "    Dense(32, input_shape=(X_train.shape[1],), activation='tanh'),\n",
    "    Dropout(0.2),\n",
    "    Dense(32, activation='tanh'),\n",
    "    Dense(y_train.shape[1], activation='linear')\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_4 (Dense)              (None, 32)                6752      \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 32)                0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 32)                1056      \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 33)                1089      \n",
      "=================================================================\n",
      "Total params: 8,897\n",
      "Trainable params: 8,897\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "nn.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "nn.compile(Adam(1e-3), 'mse')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 47.9 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "h = nn.fit(X_train, y_train, 1024, epochs=3000, validation_data=(X_valid, y_valid), verbose=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x20c4279e3c8>]"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYQAAAD8CAYAAAB3u9PLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x20c41496208>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(np.sqrt(h.history['loss'][500:]))\n",
    "plt.plot(np.sqrt(h.history['val_loss'][500:]))\n",
    "#plt.yscale('log')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.16069505"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rmse(y_valid, nn.predict(X_valid, 10000))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Early stopping\n",
    "\n",
    "From the figure above we can see that the validation score reaches a minimum. Ideally we would like to stop then."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from keras.callbacks import EarlyStopping"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "es = EarlyStopping(patience=200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "nn = Sequential([\n",
    "    Dense(32, input_shape=(X_train.shape[1],), activation='tanh'),\n",
    "    Dropout(0.2),\n",
    "    Dense(32, activation='tanh'),\n",
    "    Dense(y_train.shape[1], activation='linear')\n",
    "])\n",
    "nn.compile(Adam(1e-3), 'mse')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "h = nn.fit(X_train, y_train, 1024, epochs=3000, validation_data=(X_valid, y_valid), verbose=0, callbacks=[es])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.15597911"
      ]
     },
     "execution_count": 187,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rmse(y_valid, nn.predict(X_valid, 10000))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df = create_sub(nn.predict(X_test, 10000), 'nn_es.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Add recent observation data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "mean, std = Xobs_train.mean(0), Xobs_train.std(0)\n",
    "Xobs_train = (Xobs_train - mean) / std\n",
    "Xobs_valid = (Xobs_valid - mean) / std\n",
    "Xobs_test = (Xobs_test - mean) / std"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5014"
      ]
     },
     "execution_count": 198,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.isnan(Xobs_train).sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 193,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((2507, 210), (2507, 24))"
      ]
     },
     "execution_count": 193,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape, Xobs_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "Xconcat_train = np.hstack([X_train, Xobs_train])\n",
    "Xconcat_valid = np.hstack([X_valid, Xobs_valid])\n",
    "Xconcat_test = np.hstack([X_test, Xobs_test])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 209,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearRegression(copy_X=True, fit_intercept=True, n_jobs=1, normalize=False)"
      ]
     },
     "execution_count": 209,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr_obs = LinearRegression()\n",
    "lr_obs.fit(Xconcat_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 210,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train R2 =  0.5844313282686324 , Valid R2 =  0.46287938500419035 , Valid RMSE =  0.16266052\n"
     ]
    }
   ],
   "source": [
    "print_scores(lr_obs, Xconcat_train, Xconcat_valid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 211,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAY8AAAEKCAYAAADq59mMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzs3Xd4lFXa+PHvmZ7eE0hCCF1AesCCdEUUFetiF9BXxVdZ1t8W69p91bXtuuvae8GKsru6LCrYUUCRHgik9zZJJpNMPb8/JmYDRCCQySTh/lzXXPO088z9BJ17zvOcorTWCCGEEB1hCHUAQggheh5JHkIIITpMkocQQogOk+QhhBCiwyR5CCGE6DBJHkIIITpMkocQQogOC2ryUErNUUplK6VylFI3tbP/RqXUNqXUJqXUp0qp/i3bxyqlvlVKbW3ZNz+YcQohhOgYFaxOgkopI7ATOAUoAtYBF2mtt7U5ZgbwndbaqZRaDEzXWs9XSg0FtNZ6l1IqFdgADNda24MSrBBCiA4xBfHck4AcrfUeAKXUMmAe0Jo8tNar2xy/Fri0ZfvONseUKKUqgCTgF5NHYmKizszM7Mz4hRCi19uwYUOV1jqpo+WCmTzSgMI260XAcQc4/krg4303KqUmARZgdzv7rgauBsjIyGD9+vVHEq8QQhx1lFL5h1MumM88VDvb2r1HppS6FMgC/rTP9r7Aq8BCrbV/v5Np/YzWOktrnZWU1OHEKYQQ4jAFs+ZRBPRrs54OlOx7kFLqZOBWYJrW2tVmezTwL+A2rfXaIMYphBCig4JZ81gHDFFKDVBKWYALgRVtD1BKjQOeBs7SWle02W4BlgOvaK3fCWKMQgghDkPQah5aa69S6npgJWAEXtBab1VK3Q2s11qvIHCbKhJ4RykFUKC1Pgv4FTAVSFBKLWg55QKt9caOxODxeCgqKqK5ublzLkockM1mIz09HbPZHOpQhBBBFrSmul0tKytL7/vAPDc3l6ioKBISEmhJTiJItNZUV1fT0NDAgAEDQh2OEOIQKaU2aK2zOlquV/cwb25ulsTRRZRSJCQkSC1PiKNEr04egCSOLiR/ayGOHr0+eQghRG/W8M/3sT/3WJd/riSPIIuMjDyk4+6///7DOv9VV13Ftm3bDn6gEKJX8ezeTNGvZlH021uxv/Y82ufr0s+X5BECvnb+kX8peWit8fv36x/Z6rnnnmPEiBGdFpsQonvTNQXU3HIBe+adj2NrMclzBtD/zbdRRmOXxiHJo4usWbOGGTNmcPHFFzNq1Ki99t100000NTUxduxYLrnkEvLy8hg+fDjXXXcd48ePp7CwkMWLF5OVlcXIkSO54447WstOnz69dViWyMhIbr31VsaMGcPxxx9PeXl5l16jECKIavNpfvpK8s6cSfn7WwgbkMjAZc+T8PhHqL5d/wMymD3Mu5W7/rGVbSX1nXrOEanR3HHmyEM+/vvvv2fLli37NWV94IEH+Otf/8rGjYFuLHl5eWRnZ/Piiy/y5JNPAnDfffcRHx+Pz+dj1qxZbNq0idGjR+91nsbGRo4//njuu+8+fv/73/Pss89y2223HeFVCiFCqno3/tV/ouqNj6jeHo4xPILUO5cSPX9BSBupHDXJozuYNGnSIfeB6N+/P8cff3zr+ttvv80zzzyD1+ultLSUbdu27Zc8LBYLZ5xxBgATJkxg1apVnRe8EKLreF2Q+yVsegvHqhWUrY/B44gg5oxTSb71DkxxcaGO8OhJHh2pIQRLRETEYR2bm5vLww8/zLp164iLi2PBggXt9qcwm82tv0SMRiNer/fIgxZCdA1nDexaBdn/Qu/6lKZSF7W7Y6jPjceSkU7GX+8l4vgDDUzetY6a5NHdmc1mPB5Pu0N71NfXExERQUxMDOXl5Xz88cdMnz6964MUQnSumlzI/hiyP0LnfkNjuYmG8jgcxUl4610os5nE664i4ZprMFitoY52L5I8uomrr76a0aNHM378eO6777699o0ZM4Zx48YxcuRIBg4cyOTJk0MUpRDisDlroHJHyysbcr/AX7KdxjIrDZUpNBSk42/yoMLCiJwyhahTTiFy+jSMUVGhjrxdvXpsq+3btzN8+PAQRXR0kr+5OOo5KqFiWyBBVGUH3it3QGMl2geuehNN9kgaa5Nw5LnQbi+GmBiiZswg6pSTiZg8GYPN1mXhHu7YVlLzEEKIzlC+DT5/ALZ9CID2g6splmZ3Gk11mTSXp+EqqkF7As8iTckRxJ5/FlGnnEJ4Vhaqh41GLclDCCGORGU2rHkAz/oPaaqNxqmm0VzupTm3BO1yAbUYIj3YRo4kbuYZhI0cie3YYzH369ejx4OT5CGEEB2ktcaz6Uucb/0J54+bcFZZ8TSkAKDCirCNHEHchVOxHXsstmNHYunfH2XoXX2yJXkIIcQhcBcU4PjyS5q++Rzn99/hbXADYIyIJSxrEnHHTyY8awK24cNRpt7/1dr7r1AIIQ6T9nppWL0a+5vLaPzmGwBM4T7Ck32En3Yc4ectxTIqq9fVKg6FJA8hhNiHp7wc+9vvYH/nHbwVFZhirCSOchAz0It5+gLUSUshKiXUYYZUUNOlUmqOUipbKZWjlLqpnf03KqW2KaU2KaU+VUr1b7PvCqXUrpbXFcGMM5gOdUj2jsrMzKSqqgqAE088sd1jFixYwLvvvhuUzxeit9F+P46vvqbohhvImTmLqiefxBrnI32qncGnl5C06GIst/6AOu3/jvrEAUGseSiljMDfgFOAImCdUmqF1rrt5BM/Allaa6dSajHwEDBfKRUP3AFkARrY0FK2NljxdiWfz4exE4dP/qalOi2EOHTa48FTVoanqIimLVuwv/sunvwCjLExJEzpR2z0eixRlTBhAZz0G4hODXXI3Uowb1tNAnK01nsAlFLLgHlAa/LQWq9uc/xa4NKW5VOBVVrrmpayq4A5wJtBjDeo1qxZw1133UXfvn3ZuHHjXhM4/f3vfyc3N5eHHnoIgJdeeokNGzbwxBNPcPbZZ1NYWEhzczO//vWvufrqq/c7d2RkJA6HA601N9xwA5999hkDBgygt3QAFeJw+ZuacGVn4y4qxlNUiLuoCE9RMZ7CQjxlZdBmbp2wMceSNHk0Ub5PMZgUjL8cTroRYtJCeAXdVzCTRxpQ2Ga9CDjQqF5XAh8foOx+/4JKqauBqwEyMjIOHM3HN0HZ5oPF3DF9RsFpDxzy4b80JPv555/PCSec0Jo83nrrLW699VYAXnjhBeLj42lqamLixImcd955JCQktHv+5cuXk52dzebNmykvL2fEiBEsWrToMC9OiJ6teedOCq+9Fm9Jaes2Y0ICltQ+hI0YRPSUsVjibZhjTVhUKea890FrmHgZTPl/EJMewui7v2Amj/Z6v7T7U1gpdSmBW1TTOlJWa/0M8AwEhic5vDC7zi8NyZ6UlMTAgQNZu3YtQ4YMITs7u3X8qr/85S8sX74cgMLCQnbt2vWLyeOLL77goosuwmg0kpqaysyZM4N3MUJ0Y43ffkvRDUswmPykzUvGGuHAbKrF4NkKus2PyJqWl8EM4y4JJI3Yg/wQFUBwk0cR0K/NejpQsu9BSqmTgVuBaVprV5uy0/cpu+aIoulADSFYDjQk+/z583n77bc55phjOOecc1BKsWbNGj755BO+/fZbwsPDmT59ertDsbfVk3usCtEZ7O8vp/SPt2ONN9PvuDzMg8dB9AiISITwBAhP/O9yRGJgPTwBTJZQh96jBDN5rAOGKKUGAMXAhcDFbQ9QSo0DngbmaK0r2uxaCdyvlPp5xpPZwM1BjDXkzj33XO677z769+/Pgw8+CEBdXR1xcXGEh4ezY8cO1q5de8BzTJ06laeffprLL7+ciooKVq9ezcUXX3zAMkL0Flprqp74K1VPPklEfwtpE/IwnnU/HL841KH1SkFLHlprr1LqegKJwAi8oLXeqpS6G1ivtV4B/AmIBN5p+cVcoLU+S2tdo5S6h0ACArj754fnvVVcXBwjRoxg27ZtTJo0CYA5c+bw1FNPMXr0aIYNG7bXzILtOeecc/jss88YNWoUQ4cOZdq0aQc8XojeQrvdlN5+O3UfriBmmKJvVgXqgtfgmLmhDq3XkiHZRaeSv7noar76eopuWILzu+9IGusiYYIJdclbkDou1KH1CDIkuxDiqOMpLqbgmmtw5+aSekIdMRP7w8VvQ2y/gxcWR0SShxCiR2ravIXCxYvRjXVkTK0g4sST4IKXwRYd6tCOCpI8hBDdnr+xEdfu3bh25eDatQtXTg7O9esx2fz0m16MdealcPrDYOxZEyr1ZJI8hBDdiqe8AufabwNJYlcOrpwcPMXFrfuV1YIlNZHowQaSB+VhOutOOHEJSDP1LiXJQwjRbXjKysg973x81dVgNmPN7E/Y4FRiJ/bFaq3B6s/BrPJQhjywRMG8F2Hk2aEO+6gkyUMI0S1ot5viXy9FOx30v24iYYbdqOqvApOBA8QPhPRpkJYF6VmQcqx07AshSR5CiG6h/MGHaPrpJ9JOqiO8cQ2kT4CRcwPJIm0CRLQ/LI8IjaNv+qsu1hnzeeTl5XHsscceVtk1a9Yc1pDt69evZ8mSJYf1mUJ0VN0//kHt668TP9JH9Ji+sPQnuGw5zLgFhs6WxNENSc0jBDp7Po8DWbNmDZGRke1OGOX1ejH9wlzLWVlZZGV1uN+QEB3WvHMnpbf/kbBUE8kTGuGijyAs7uAFRUgdNcnjwe8fZEfNjk495zHxx/CHSX84pGMPNJ9HW48++igvvPACAFdddRVLly4FAl/0V1xxBT/++CNDhw7llVdeITw8nJtuuokVK1ZgMpmYPXs2Dz/8cOu58vLyeOqppzAajbz22ms88cQTPP/888THx/Pjjz8yfvx45s+fz9KlS2lqaiIsLIwXX3yRYcOGsWbNGh5++GH++c9/cuedd1JQUMCePXsoKChg6dKlUisRncLX0EDxDUswmLykTSpFzX8bEgeHOixxCI6a5NEd/NJ8Hj/bsGEDL774It999x1aa4477jimTZtGXFwc2dnZPP/880yePJlFixbx5JNPsmjRIpYvX86OHTtQSmG32/c6X2ZmJtdeey2RkZH89re/BeD5559n586dfPLJJxiNRurr6/niiy8wmUx88skn3HLLLbz33nv7xbZjxw5Wr15NQ0MDw4YNY/HixZjN0qZeHD6tNSU334y7sID+0ysxn3MfDJJpBHqKoyZ5HGoNIZh+aT6Pn3311Vecc845rUO3n3vuuXz55ZecddZZ9OvXr3WOj0svvZS//OUvLF26FJvNxlVXXcXcuXM544wzDimOCy64oPW2WV1dHVdccQW7du1CKYXH42m3zNy5c7FarVitVpKTkykvLyc9XSbLEYev+rnncHzyKSnj6gg/7WKYtP8smaL7kgfmXehA83kAB5w2dt95OpRSmEwmvv/+e8477zw++OAD5syZ0+E4br/9dmbMmMGWLVv4xz/+8YvzhVit1tZlo9GI1+s9pM8Soj2Na9dS+dhjRPd3ETd7XKB3uHTy61EkeXQjU6dO5YMPPsDpdNLY2Mjy5cuZMmUKAAUFBXz77bcAvPnmm5x00kk4HA7q6uo4/fTTefzxx9m4ceN+54yKiqKhoeEXP7Ouro60tMAMvy+99FLnX5QQ+/CUlVH8m99gifbT95RI1PxXZViRHkiSRzcyfvx4FixYwKRJkzjuuOO46qqrGDcuMKz08OHDefnllxk9ejQ1NTUsXryYhoYGzjjjDEaPHs20adN47LHH9jvnmWeeyfLlyxk7dixffvnlfvt///vfc/PNNzN58mR8Pl/Qr1Ec3bTbTfGSJWhHHenTnBgufwvC40MdljgMMp+H6FTyNxcHUnb3PdS+8QZpk2uJ/sPLMPTUUId01Dvc+Tyk5iGE6BLVL7xI7RtvED/MQfSVt0ri6OGOmtZW3Ul1dTWzZs3ab/unn35KQoL0pBW9i9aaykcfpfrZ54jq10Ty5XPhhOtDHZY4QkFNHkqpOcCfCcxh/pzW+oF99k8FHgdGAxdqrd9ts+8hYC6B2tEq4Ne6l9xjS0hIaPfhthC9jfb5KLvjDuzvvkfsoEb6XHISat6fpWVVLxC021ZKKSPwN+A0YARwkVJqxD6HFQALgDf2KXsiMJlAUjkWmAhMC1asQojO53e7KV5yPfZ33yNhRAN9li5CXfgamKwHLyy6vWDWPCYBOVrrPQBKqWXAPKB1XA6tdV7LPv8+ZTVgAyyAAsxAeRBjFUJ0Ip+jkaJrr8S5/ieSxztI+P1DMPbiUIclOlEwk0caUNhmvQg47lAKaq2/VUqtBkoJJI+/aq2373ucUupq4GqAjIyMIw5YCHHkvLW1FF5xMc27cuk7xUPsrcsgc3KowxKdLJitrdq7qXlIzyyUUoOB4UA6gSQ0s+X5yN4n0/oZrXWW1jorKSnpiIIVQhw5T2kp+eediSsnl/TTw4h9cJUkjl4qmMmjCOjXZj0dKDnEsucAa7XWDq21A/gYOL6T4+sSoZ7Po6Neeuklrr8+0BLmqaee4pVXXglpPKLncO3eTd65c/FWVtLvon5E3bs6MPuf6JWCedtqHTBEKTUAKAYuBA71pmcB8D9Kqf8jUIOZRqBVVq/QlfN5HIlrr7021CGIHqLpx3UUXrkIfC76Xz8F21VPyZAjvVzQkofW2quUuh5YSaCp7gta661KqbuB9VrrFUqpicByIA44Uyl1l9Z6JPAuMBPYTOBW17+11v84knjK7r8f1/bOnc/DOvwY+txyyyEdG4r5PPx+PwMHDmTjxo3ExsYCMHjwYL7++mu+//577r33XtxuNwkJCbz++uukpKTsFcudd97ZOpz7hg0bWLRoEeHh4Zx00kkd/luJ3slXX0/VPb+l5l9fYg7zkXHXQizzbpKmuEeBoPbz0Fp/BHy0z7Y/tlleR+B21r7lfMA1wYwtFLp6Pg+DwcC8efNYvnw5Cxcu5LvvviMzM5OUlBROOukk1q5di1KK5557joceeohHHnnkF2NfuHAhTzzxBNOmTeN3v/tdp/5dRM+jfT7qnn2AimfewOf0ETvCTNJtD2AaPy/UoYkuctT0MD/UGkIwhWI+j/nz53P33XezcOFCli1bxvz58wEoKipi/vz5lJaW4na7DxhXXV0ddrudadMCXW0uu+wyPv7448P+O4ierek/yyj7vwdoLnURlgIpt15J2Nm/AeNR83UikLGtulQo5vM44YQTyMnJobKykg8++IBzzz0XgBtuuIHrr7+ezZs38/TTT//iPB4/x7Xv54ujj3fb15TMn0zekrvw2ptJvfpU+n/yI2Hn/U4Sx1FIkkc3Eoz5PJRSnHPOOdx4440MHz68deystvN4vPzyyweMKzY2lpiYGL766isAXn/99U67ZtH96fKdVP/mdHb/ahF1m6tJmDOagZ9+TsyNj6PMtlCHJ0JEfi50I23n8wBa5/PIy8trnc/jmmuuYciQISxevJi6ujrmzZtHc3MzWut25/OAwK2riRMn7jXZ05133skFF1xAWloaxx9/PLm5uQeM7cUXX2x9YH7qqTIaaq+mNVTnoLNX0vDP96n8pBB3g4nIY9NIufcxLMeMCXWEohuQ+TxEp5K/eQ/laYK8r2HXf9DZ/6Z+UzlVW6Nw15uxpESRctPNRJ52TqijFEFwuPN5SM1DiKNVfQlkfwS7VsGez9HuJuqLY6jaEY+7Oh7LgAxS71hC9Jw5qB7QL0l0LUkeISDzeYiQqt4NXz0KPy0DvxcdnUG9fwZVX5fgLq7AOiSDtNuvI2r2bJRBHouK9vX65NEdWwr11vk8esst0O7MuX49hogIbIdza7B8K3z5CGxdDkYLetwC6moHU/XaB3gKNmI95hjS/nIrUSefLElDHFSvTh42m43q6moSEhK6XQLpbbTWVFdXY7NJ65tg8JSWUnbvfTg+/RSAiMmTSbjmasInTjz4f9tFG+DLhwO3qCyR+MZeg70kldrHV+Ap/gjriOGk//UJImfOlKQhDlmvfmDu8XgoKio6YB8G0XlsNhvp6emYzTKmUWfRPh+1r79B5eOPo71ukiZa0D4XNZvB1+gjbFAfEuafTuSpc1HxA8Ac9t/CeV/DF3+CPavBFour34XUbFPUffQfdFMT4RMnEr9wIZEzpsuPq6PY4T4w79XJQ4ierHnHDkpvvYXmrduJSNf0GVuBZfAISBiCv2I39u+LqNlixNNowhrjIWG4g+iRcajEAeB1QfF6dFgSjvAzqFlfg/O7dSirlZizziTu0kuxDRsW6ksU3YC0thKil/A7nVQ+8gA1b76L0eIj9YQ6omccj5r8dxg4HZTCAMRfp4mrr6T+/Tepev09StZWUrnTRsIEO1H9NXXMp3blbjwlKzH17UvS/7uR2PPPxxQXF+IrFL2B1DyE6EYcH75K2QOP4Kl1ETOwiZSLZ2A8+UboM+qA5bTfj+Ozz6h65lmaN21q3R6elUXcZZcRNWsmyiS/FcX+pOYhRA/my/mesj/cQP3WeizRPvrfMJ3wS26H2H4HLwwog4Gok08mctYsnGvX0vj990TPnn14rbKEOASSPIQIJa3xffk0BTc9QrPdSOKZE0i47VEMMSkHL9sOpRQRJ5xAxAkndHKgQuxNkocQodJch+/d6yn8+zc02y2kP3w/UafLECCiZ5DkIUQoFG/A/+ZCij5spKnWStrjjxM1e3aooxLikEmPICG6ktbw7d/wP3MqRR97cFZbSXv4YaIlcYgeJqjJQyk1RymVrZTKUUrd1M7+qUqpH5RSXqXU+fvsy1BK/UcptV0ptU0plRnMWIUIusZqePNC/B/dQtEPg2gsUfS9/36iTz891JEJ0WFBSx5KKSPwN+A0YARwkVJqxD6HFQALgDfaOcUrwJ+01sOBSUBFsGIVIujyv4GnTkLv+ozinJNo3FVPn7vuJPbss0MdmejBtNZ8vrOSDzcWd/lnB/OZxyQgR2u9B0AptQyYB2z7+QCtdV7LPn/bgi1JxqS1XtVynCOIcQoRPD4PfPkofP4AOqY/xSVzcfz4HSm330bcr34V6uhED+Xza/69pYwn1+SwtaSekanRnDUmtUuHmQlm8kgDCtusFwHHHWLZoYBdKfU+MAD4BLhJa+3r3BCFCKLyrfDBYij9CT3yfErWRtPw+b9J/sMfiL/kklBHJ3ogt9fP8h+LePrzPeypaiQz0cKFM0sYnlqDUlO6NJZgJo/2UuChdmc3AVOAcQRubb1F4PbW83t9gFJXA1cDZGRkHG6cQnQunwe+ehw+fxBsMejzXqT0rR+o/9d7JC1dSsLCBaGOUPQwTreXN78v5Nkv9lBW38wxaUbOm7mDDbX/4F+l1VTqSVw+8tJeU/MoAtp2j00HSjpQ9sc2t7w+AI5nn+ShtX4GeAYCw5McacBCHLGyLfDhdVD6E4w8F+8Jt1Dx1xeo++ADEq+7jsRrrwl1hKIHsTvdvPxNPi99k0ut08PYgX6yJqzju6qP+U9pM5PTJrNw5EIm9ZnU5SMjBzN5rAOGKKUGAMXAhcDFHSgbp5RK0lpXAjMBGbhKdF8+D3z1GHz+EITF4j31SWq+LqXm/vlol4uExdeSeMP1oY5S9AANzR5WZ1eycmsZn22voMnj47hjGolM/or1VZ+TX2Hg9AGnc8XIKxgaNzRkcQYteWitvUqp64GVgBF4QWu9VSl1N7Bea71CKTURWA7EAWcqpe7SWo/UWvuUUr8FPlWBdLoBeDZYsQpxRMo2wwfXQdkmvJlnUV0ylNrrHkG7XETPnUvi4muxDhwY6ihFN1blcLFqWzkrt5bxTU41bp+fhEgjJ4wqw2H7lG21G4m0R3LFyCu45JhLSIk4vOFrOpOMqivE4Wqqhe+egS8ewksc1c6TqV35fSBpnDGXxGsXYx04INRRim6qsMbJyq1l/GdrOevza/BrSEt0M2xAMX7bdnbUbaDR00ifiD5cOvxSzhtyHpGWyE6PQ0bVFSKYtIa6QihYCwXfBt4rtuFtNlBdPYHaddVo95eSNMQv8vr8bCy089mOClZnV7K9tB7wMyCtmhOz8qhTm8lr2Mk6B6T4UzhtwGlMSZvClPQpmA3db3ZOSR5CtMfvg4pteyULbS/G7TDR3BCFy5tOsz0LZ24V2l0qSUO0q9rh4vOdlazOruSLnZXUNXkwmhsZnFHMhKwcSt0bqfLUU9toZEzSGM4espQp6VMYEjuk208NLMlDiH3lfYX/3WtxFZTRXGum2RGLyxFJc3kG2u0FQJnrsQ4ZQsw5U4m/7HJJGgII1C62ldazekclq7Mr+KnIjtY+4hNKyBxcgMu8neKmHEqAeF88MzKmMyV9Cif0PYEYa0yow+8QSR5C/MzrhjX343jn75R8H4+vKQkAQ2QktmOOIXbGcGzHDMc2YjjWgQNRFkuIAxahpLWmqLaJn4rsbCyw81ORnc3FdTR7/Bgs1WSmFzFy7G4qPFto8jkp8JkYHT+a845ZwolpJzI8fjgG1XPHppXkIQRA1S7024uo+E8eNTvisQ4eRJ/rb8A2cgTmtDSUoef+Ty46R0Ozhx8L7PxUaGdjYSBZVDncgA9bRAVpfSoZPKIYB7uodpdQCaSZ0jgjYy6T0yZzXJ/jgvLAO1QkeYijm9aw4UXc79xG8ddRNFdFEnvhfFJuugmDzRbq6ESI1Tk9rNpezsebS/lyVxVunx+DuY60lAr6DiglyZxPhSsHt99FBRBviGd04miOT72cE1NPJDM6s9s/uzhckjzE0auxClbcQP2qzyjdkAjmMNL+fD/Rp8rcGkez2kY3q7aV89GWUr7OqcKjm0lOzmXIsbuoI5s6dxV2oNFvZnjkcKZnXsDopNGMShxFWmRar00W+5LkIY5Ouz7B/951lH/pwb47nrAxo0h95BEs6WmhjkyEQE2jm5Vby/hocynf7q7GpxwkpeQw8NhsKr1baPK7aVBxTE49gdFJoxmdOJph8cOwGI/e516SPMTRxdUAn91L88rnKP6+D+5aEwn/cxVJS25AmbtfW3oRHF6fn5+K6vhiZyVf7qpkY6EdbbST3GcXmcfuoMKznSb8xJr7Mn/Qr5iVMYtxyeMwGoyhDr3bOGDyUEp9Z1peAAAgAElEQVTN1Fp/1rI8QGud22bfuVrr94MdoBBHTOtAX40fX0NvWY59B5T/1AdDdBz9nnuQyMmTQx2h6ALF9qbWZPHVrirq3Q5M4fn0TSmh/6gcqjy7cQJ9wwdxVsaVnNz/ZIbHDz9qbkN11MFqHg8D41uW32uzDHAbIMlDdF/1pegfX8f16Ws4cypwVofjrE7G53ATMfkEUh98AFNiYqijFJ3M79dUOVwU25soqm3ih4JavthZye7qKozhuUTHFhI9KA/8+Wj8OJSJY2OP5ZJ+v2ZWxiwGxEifnUNxsOShfmG5vXUhOo3WGt3cjL+xEX9jI9rtRoWFYbDZMISFoWw2lHH/Wwi6qZHmlS/iXPUezu15OCst+D0GIAZzal8iZ00kYvJkos84Q5rf9mDNHh8/FtgprHVSYm+iuLaJYnsTJfYmSuxOPKoOg7kWZa7FEl5MZEI+UYlFgEYZzAxKHEVWn1PISsliTNIYws3hob6kHudgyUP/wnJ760IclPZ4cOfn48rZjSsnB/ee3XhraluThL+xEb/Tib+xEfz+A55LWa0YbDZUmA2DUaN0M+6KOrQ38LvGkpJM9OknEn7STMInTMCcmtoVlyiCpMTeFBgXakcFX+8uw2MqQJlrMFhqCQ+vx2KrQyfXEJZcjZX/TjpqNdoYnTSGCX3OJCsli1GJo7CZpBn2kTpY8hiolFpBoJbx8zIt61K3Ewfkys3FtXMXrpxdgUSRk4MrLx88nsABCszRBkwRRoyR0ZgTEzEMHYYhIQ1DVDSGiAgMEREYIyJQFgv+Zhf+Jie6sRF/5R78ZTnoynz8dXn4PaD9RsKz0gmffgbhp1+OKTn0w1aLw+fzazYW1vLZjgo+3V7BjjI7xojdxCZtI3zIZjy6sfXY+LAk+kb2JS1iAqmRqXu9+kX2w2yUxhCd7WDJY16b5Yf32bfvuhCtqp59lspHHm1dN6enYx08mMiskVj9OVjqv8Ua0Yih/wQwh0HpJnBtDhzsN4LlGEgcDX3HQJ/+EBYHeV/BnjVQ9SW46yFewcjRMPAKGDgdMo4PnEv0WA6Xl8+zK/l0ezmrsyuodTZjjsylT99skhI20uxvwGyOYGa/mczqP4tBMYPoG9kXq9Ea6tCPOgdMHlrrz9uuK6XMwLFAsda6IpiBiZ7LW11N9d+fImLKFJKWLMHaPx1D3ipY9ywUfhf4gp9xAUy8KpAcINAiyp4fmL61dFPgPedT+OnNvU8emwEjz4FBMyBzKkQkdP0Fik5V7XDxyfZyVm4t56ucKtxeLzFxhfQdmI2FDTh9dbhN4czsN505mXM4Me1ESRbdwMGa6j4FPNEyA2AM8C3gA+KVUr/VWr95oPLi6FT11NP4XS5SbliIteJ9+Pcr0FgJ8QPh1Pth7MWBmkRbSkFcZuA1ok2Ft6EskEgaq6D/iRAvd0t7g6JaJyu3BmbOW59Xg1/7SUkuYcSx2VT419PgqaVGhTE9fTqnZp7K5LTJ8pyimznYbaspWutrW5YXAju11mcrpfoAHwOSPMRe3EVF1C5bRuy00ViXnx6oUQydA5OugoEzoaMtnKL6BF6ix/L5NUW1TnZXOthcVM9/tpWxtSQwEVJmWjkTxu+g1LuOOncNJb4wpqRNYXbmbKamTyXMJLchu6uDJQ93m+VTgHcAtNZlh9JxRik1B/gzgTnMn9NaP7DP/qnA48Bo4EKt9bv77I8GtgPLtdbXH/QDRchVPfEEyqBITPgG0ifBuc9AXP9QhyW6QEOzhz2VjeypcrC7opHdlQ52VzrIq3Li9v3ccs7PMZmVnDhpB4Xu76h21eB0/zdhTEmbIs1me4iDJQ+7UuoMoBiYDFwJoJQyAQf8SaCUMgJ/I5B0ioB1SqkVWuttbQ4rABYAv/2F09wDfP4L+0Q305y9k7oV/yBhWn/MpkI483FJHL2I1poqh5uCmkbyq50U1DgpaHnPr3FS2eBqOdKPydJA3/gmEuIbyUqvw2CpxU01ZU15FLtqqHbamJI+hVMzT5WE0UMdLHlcA/wF6AMs1VqXtWyfBfzrIGUnATla6z0ASqllBFpvtSYPrXVey779GvQrpSYAKcC/gQ5Pzi66XuVjj2GICCchfi0cdy0kDw91SOIwVTtcbCqq46ciO9tK6gOJosaJ0+1BGZ0oUx0Gcz2xUU6iIp0kZjpIMNlx6Urs7gq82osdsAM4IVEnkhqZyknxk5nabypT06ZKwujhDtbaaicwp53tK4GVBzl3GlDYZr0IOO5QglJKGYBHgMsIJCrRzTk3bMCxZg1JU2MwxiXA9JtCHZI4RA3NHjYX17GpqI5NRXZ+Kqyj2N6AwVaCKSKP6JgyzHH1RCXUYfbX4sfbWtYFuFEkmBPoG9GXtMjRpEamkhaZ1vreN6KvPOzuhQ7W2uovB9qvtV5yoOLtFTmUoIDrgI+01oUHerailLoauBogIyPjEE8tOpvWmopHHsUUF0l88g6Y9QTYetZ8zL2dx+en1N5MYa2TwpZaRGFtE9tK6thT1YhWLoxh+cTFF2NLzScudQ9eHXjkmdiSCJLDh5EcnkxKeArJ4cmtywlhCZgN0gnvaHOw21bXAluAt4ESOjaeVRHQr816ess5DsUJwBSl1HVAJGBRSjm01nv9nNVaPwM8A5CVlSXDpYSIY80amn74gT6T/RgyxsHYS0Id0lHL5fXxQ76dHwpqW59HFNY6Ka1z4qMJZWpEGZ2YTE3ERXmISigns+8eqj15aPx4lIFB8ccwPvlXjE8Zz7jkcSSGyeCRYn8HSx59gQuA+YAXeAt4T2tdewjnXgcMUUoNIPDA/ULg4kMJSmvd+u2jlFoAZO2bOET3oH0+Kh99DHNSJLFpO+H01zveHFccNp9fs7Wkjq9zqvlydyE/lv+Ez5KLwVaKxeLEZG6CPo1E9HGi2fvRohPwG22MThjN+SmzGZc8jjFJY4gwR4TmYkSPcrBnHtXAU8BTSqk04CJgq1LqD1rrVw9S1quUup7AsxEj8EJLZ8O7gfVa6xVKqYnAciAOOFMpdZfWemQnXJfoIvX//CeuXbtIm1yPGn8ppEvbhmDSWrOnqpGvd1XyaU42P1ZsxGXcjTE8H6O1DFOaxoyBzOgBJEekEGuN3esVY43Za71PRB8Z90kcFqX1we/2KKXGE0gcpwAbgEf2aXIbcllZWXr9+vWhDqPb8Tkc2N9+h7rl72MdOozEa6/BOmRIp5zb73az57TTMfqqyZxdhVqyASKTO+XcRzun20tuVSN5VU5yqxxsryxmtz2H0qZcXMZcjGH5GMz1AFgMYRybOIrj+k5gbPJYqT2IDlFKbdBad/hX38EemN8FnEGgo94y4GattfdAZUT34KmooPbVV6ld9hb+hgZso0fTsHo19f/6F1GzZ5O4+Fpsw4+sKa39rbfxFBfTZ1o1asYdkjha+P2akrqmwBd/dSP5VY04PYEhwn9+aKgUqJa1wDI0e/zsrqkgr343dd4iDLYyDNYyjNZylLEJbIANEkxJjE0+jsnpWYxLHseQuCGYDDKjtOhaB6x5tPS/2AM0tWz6+WAFaK316OCGd+ik5hHg2rOH6hdeoP7DFWifj6jZs0lYtJCwsAq8Dic1n2yl9q338TscRM6YQeJ1iwkbNarDn+NzNLJ79ilYbXYyzolAXfcNHAW3Pzw+P40uLw3NXhwuL7WNbnKrG8mramR3VS177HmUNRXgN1VgsFRhsFRisFahlAdQoI0t7wbACFoBhsBLecHY0PpZNmME/aMGMiJhKMckDGVI3BAGxQ4i3hYfmosXvVJQah7InB09gtaaph9+oPr5F3B89hnKaiX2gvOJX7AAS5SGfyyBPWswAclAwtlJ1OT3oea7r8hbvZqIiWNIvGEp4ZOOP+TPrHn5JXw1tSSfUok6/Z0uSRw+v8budFPd6Kba4aa60dXy7qba4aKm0Y3Hd+AJpCAw3NZ+2/Y7RtPs8dPg8tDgduBw19Poq8ejA62VlNGJMjhRJgcGSyVGaxXKbIdYMMcGzhFvTSEzOpOh8VOJskTi0z58fl/gXfvwaz9evxe/9uPTPozKyMCYgQyOG8zg2MGkhKfI/Nmi2zrYA/P89ra3DD1yIdDuftF1mrZupfyee2nauBFjbCyJ//u/xF1yMabYWFj/PLx6R+C+yOkPQ9IwKN+KsXwLSclbie9bTu0OEzWbfiD/8oWEpxuJnz4My8TZmCfOwxDf/oCE3poaap5/nqgMN2EnzQkMj96JPD4/uysdbCmuZ2tJHVtL6tldWYfdVYM2OlBGB8rUgMH087IDi7URo6kRlB/V8kte6ZZf9Bj22aZA+VtaH/nR+Nqs+wLvyoe2uPBZAucEMLe82gozRpAelcGQuMkMiMkkMyaTAdEDyIjOkEH9RK92sGce0cD/EugtvgJYBVxPYCyqjcDrwQ5Q/DJXbi6FV16FslhIuf02Ys89F0NYGFTvhpcvhfyvYdBMOPPPgXkwAAZMbS1v9PtIrNlDfP4P1L6/guqVmyh6bRu8tg14HIPVgDkxDlO/AZjS+2NO6YOpTwrO79fhb2oiaYwTZt93RNfQ6PKys7yBLSX1bGtJFDvK6vEYKjCG5WONKMAWVYi7Xxnh7fQxtRnDSAxLICEsgXhbOmaDufWXvc/va/1V3/ZXv9Yao8GIURkxGoyYlKl13WQwYVRGDMpApDmSGGvMf1+WmP3WpaWSOFod7JnHh0AtgXk8ZhFoUmsBfq213tglER6io+2Zh7eykryLLsbvdJL55htY+vcHvw++ewo+vQeMFphzf6DDXptbH1pr6pu81Djd1DS6qW0MvNc43dTZHVh2biW1fCMZlZtJtudhbWzC4zTS3GRBN/33gW/swEbWTJ/LquRFxEdYSIiwEB9hIa7Nss1spMrhoqLeRUWDi8oGFxUNzVQ0uKhqCGxzuLyg3BhtRUTEFBEZU4zLuAe3Dtz7j7ZEMyZpDCMTR5IUlkRCWAIJtoTWdxkfSYgjc7jPPA6WPDZrrUe1LBuBKiBDa93wi4VCpDsmD601jW4flW2+OH9etjd5cHn8uLw+XF5/4OVps+z14fb6sZgMRFhMhFmMRFiMhFtNxGg3c1+4h+jKYn78zf00DxqG1Z7Dydl3k9a4hW2RJ/Ja0lJK/XE43T6aPD6cbh92p4dapxufv/1/c6vJQEyYGZtFYTE3Y7I4SFK59PP/RJxvF/iqcLoMuJuN1Caa2BE+DZcnGldzBA5nGG53JNobhfZGgra0nNWPMjlQpnrCbI1ERzZhszkwWRrAVI+HWmq9hfh1oDVSZnQmY5PHMi55HGOTxpIZk4lBSadDIYIlWA/MPT8vaK19Sqnc7pg4uprH56fa4d4rGVQ2uKhs+ZVd6fjvtqaWJpptmQyK2HAzVpMRq9kQeDcZsJoMRIeZW5ctRgNunx+n20ejy0uVw01TVQPXfPRXYsryuOeERazP9nJ1zkMsNb2HEys3G5bwuXsaYVUmwi1uwixG4iMspMcZiQmzEB9hJj7CSnyEmXCrj0ZdRI2ngNKmXHLrcyioL6C6uRqvP9Aiu5jA/UmAMGMSiUYb8V4vzWEx+H07aGiuQds05ti9nwfYjOGYlY1Gbx1+/vs3cACNKBJsCSSFJZEUnsGQ2FmMSx7H6KTRxNn2mWFQCNEtHSx5jFFK1bcsKyCsZf3nprrRQY2uC7iaHDjeuY6cof9DmW0gDc3elpentTlmQ7OH+mYvdU4PlS2tetoTG24mMdJKcpSVcRmxJEdZSfr5FWlrXY4NM2MwdLwVjdaa0ptupq50B33vu5dlM7PQ712JoWgdevg8rHMf5v/a6Wvh8XkoaChgl30Xu2p38VPtLnaV76LIUdR6TJgpjMGxgzmu73EtX+qBW0SJtkSSwpNIDEsk3BS+X+sfr9+L3WWnqqlqv1eTt4nEsESSw5JJCk8iOTy59daT9EsQomc7WGsrY1cFEiqOmjI8u79g4O7V3OK+nd06DQCjQRFlMxFlMxFpNRNlM5GREE5WZhxJUVaSo/6bDJKirCRGWrCagvvnqnz8z9R9+CGJS24gdrgZnp6C0hrOex416nwcbge5lZvZU7eH3Lrc1veihiK8LX07jcpI/+j+jEwcybzB8xgSN4ShsUNJi0o7rNtDJoOJxLBEGTxPiKPMIQ1P0hMc7jMPr8/Pjz98z9jPLkEpI/b5HxLedwhhZmO3amNf88YblN99D5HnnoVjUjXFO/9JYUJ/igZNJd9dS649l4qmitbjTQYT/aP6MyBmQOtrSNwQBsQMwGq0hvBKhBDdSVAemPckR/zAvHwbvDQXzOGw8KOQT5+qtWZN4Rq2VG/B8MU6Zjy1jk1DzPzfOX78bW55xVpj6RfVrzVBDIwZyMCYgaRFpckcC0KIgwrWA/OjR8oIuPxDePlMePkMWPgxxKR37BxVu+DLR2DU+TD45MMOxef38cD3D7AsexnDixS3vumlNN1KzkwHSxqN9JvwP6QPPZ30qHSiLT3+sZMQogeSmse+in+AV+ZBRCIs+Aii+x68jKcZvnoUvnoMfG5QRpj7MGQt6vDHN3mb+MMXf2B14Wqujz2b6fetxGT20H9qHqZhk+HcZw8tJiGEOASHW/OQBvT7ShsPl74Hjgp45azA+4HsXg1/PwE+fxBGzIMbfoDBs+Cfv4FVfwT/wcda+lltcy1X/ecq1hSu4dYRv2bWo2tQngb6nViIac4tgZqRJA4hRDcgyaM9/SbBxW9DXVGgFtJYvf8xjgp47yp49ezA+mUfwHnPQcIguPDNQK3j6z/DuwsDNZODKKwv5LKPLyO7JpvHTryXKY++haeikn6nguX6FTDtd2Do9Y3fhBA9hCSPX5I5GS56E2r2wKvzoKll5l2/H9Y9D09kwbYPYdofYPG3ew8OaDTB3EfhlHtg2weBGkx7CajFlqotXPrxpdib7Tw38CKyHvkd9T8WkXTqYML++DX0PzHIFyuEEB0jyeNABk6H+a9DZTa8ei7kfwsvzIZ/3Qh9R8Pib2DGLWC27V9WKZi8BC54CUo2wvMnBwYs3MfnhZ+zaOUiwpSJV5tsjFx+H2XfGAkbOZSERz6EcJm7QQjR/QQ1eSil5iilspVSOUqpm9rZP1Up9YNSyquUOr/N9rFKqW+VUluVUpuUUvODGecBDTkZfvUKlG2CF+dATS6c8zRc8Q9IPITpXEeeEzi2uQ6eOxkK1rbuejv7bZasXsIAzLyW/ROZ5bspyckCUxipf/4byii3qYQQ3VPQmuq2DKT4NwLznhcB65RSK/aZ+7wAWEBgiPe2nMDlWutdSqlUYINSaqXW2h6seA9o2GmB5xj5X8HkpR2vDWQcB1eugtcvgJfPQp/9JE+4S3h2y7NMcfl4uGw74RMWUl2QiXP7k/S9/34s6R1sJiyEEF0omP08JgE5Wus9AEqpZcA8oDV5aK3zWvbt1SRJa72zzXKJUqoCSAJCkzwAhs4OvA5XwiC46hNYdjFvrPo1zybEc169g9vCBmH6nzdptluouGk+UaecQsw5Z3de3EIIEQTBvG2VBhS2WS9q2dYhSqlJBOYQ2f+BQU8THg+XfcAPKYNJ92nuOOleTIv+gz9uGCW//z2m2Fj63H1XtxoWRQgh2hPMmkd734Ad6pGolOoLvApcobXer8OEUupq4GqAjIyMw4mx65ltFEankBk2AjXuEgAqHn0U164c+j37LKY4GZJcCNH9BbPmUQT0a7OeDpQcauGWKXD/BdymtV7b3jFa62e01lla66ykpKQjCraraK3Jr8+nf3Rg7CzH119T+8qrxF16KZFTTgpxdEIIcWiCmTzWAUOUUgOUUhbgQgLzoB9Uy/HLgVe01u8EMcYuV91cjdPrpF9UP3x2O6U334Jl0CCSf/v/Qh2aEEIcsqAlD621F7geWAlsB97WWm9VSt2tlDoLQCk1USlVBFwAPK2U2tpS/FfAVGCBUmpjy2tssGLtSoUNgcdAGZH9KL3zLry1taQ+9CAGWzt9RYQQopsK6qi6WuuPgI/22fbHNsvrCNzO2rfca8BrwYwtVPLr8wFI/TqHhn//m6QbbyRs5MgQRyWEEB0jQ7J3sYL6AlLqDHhefpKwrAkkXNnxkXeFECLUZHiSLlbYUMgFG23g9ZL6wIPSi1wI0SNJ8uhi+fX5DKgyYB02DEt6h7u9CCFEtyDJowtprSlsKCSxwoV10KBQhyOEEIdNnnl0oZrmGnR9A2F2H9bBkjyEED2X1Dy6UGFDIWkt03pYpOYhhOjBJHl0oYKGAtKrAiO0WAcPDnE0Qghx+CR5dKH8+nz6VSuUzYY5NTXU4QghxGGT5NGFCusLGVRrxjJwAMogf3ohRM8l32BdqKChgNQqjXWQ3LISQvRskjy6iNaaisp8omulma4QoueT5NFF7C470WUNANJMVwjR40ny6CJtW1pJM10hRE8nyaOLFNQXkF6twWzC0q/fwQsIIUQ3JsmjiwRqHmDJzESZpGO/EKJnk+TRRfLr8+lfY5TOgUKIXkGSRxcprc4nodYrzXSFEL2CJI8u4snLQ2lpaSWE6B2CmjyUUnOUUtlKqRyl1E3t7J+qlPpBKeVVSp2/z74rlFK7Wl5XBDPOYLM324ktdQBIHw8hRK8QtOShlDICfwNOA0YAFymlRuxzWAGwAHhjn7LxwB3AccAk4A6lVFywYg22n5vpaoMBS//+oQ5HCCGOWDBrHpOAHK31Hq21G1gGzGt7gNY6T2u9CfDvU/ZUYJXWukZrXQusAuYEMdag+rmllaFfKspiCXU4QghxxIKZPNKAwjbrRS3bgl222ymoD9Q8wocMDXUoQgjRKYKZPFQ723RnllVKXa2UWq+UWl9ZWdmh4LpSUW0efexgGzwk1KEIIUSnCGbyKALadqVOB0o6s6zW+hmtdZbWOispKemwAw22xj27MPqRZrpCiF4jmMljHTBEKTVAKWUBLgRWHGLZlcBspVRcy4Py2S3beqa8IkCa6Qoheo+gJQ+ttRe4nsCX/nbgba31VqXU3UqpswCUUhOVUkXABcDTSqmtLWVrgHsIJKB1wN0t23qcOlcd8aVOtALLgAGhDkcIITpFUAdZ0lp/BHy0z7Y/tlleR+CWVHtlXwBeCGZ8XeHnARF9fRIx2GyhDkcIITqF9DAPsp/7eJgHSq1DCNF7SPIIskJ7HqnVED1sZKhDEUKITiPJI8hqdu/A5IfwIcNCHYoQQnQaSR5B5tmzB5CWVkKI3kWSR5CZ88sAsAwYGOJIhBCi80jyCKI6Vx2J5U24EqMxRkaEOhwhhOg0kjyCqKihiLQqjc7sscNyCSFEuyR5BFG+PZe0aggbLAMiCiF6F0keQVSRuxWrFxKOGR3qUIQQolNJ8ggi565sACKGDg9xJEII0bkkeQSRf08BANZB0tJKCNG7SPIIorCiappibBhjYkIdihBCdCpJHkHS4G4gqbwZd7/kUIcihBCdTpJHkBTU55NeDcaBmaEORQghOp0kjyAp2bOZMDdEDRsR6lCEEKLTSfIIktodmwFIGTEhxJEIIUTnk+QRJK7duwGpeQgheidJHkFizC/FGWHCFB8f6lCEEKLTBTV5KKXmKKWylVI5Sqmb2tlvVUq91bL/O6VUZst2s1LqZaXUZqXUdqXUzcGMMxiiSuw40uL+f3t3H2RVfd9x/P2BBXZ5hgWSoCgotAlaFYIEmyfiQyamGfGxNU1aTNOxcaqJkowTO5M4tZN2zIO2aa0zRk1sTTQZMEpbGjAKjTGt4cEJguiAIrCE8LQsLJBlWfbbP85v63XZLXt393Dlns9rZmfPOfd37/3+9rd7v3t+55zvqXQYZma5yC15SBoI3AdcDkwHPimp8xzOZ4F9ETEVuBe4O22/DhgSEb8HvBf4i47EcipoPtLMO3cdpf3MiZUOxcwsF3nuecwGNkXE6xHRCjwOzOvUZh7wSFpeCFwiSUAAwyTVAHVAK3Agx1j71bYtaxneArVTp1Y6FDOzXOSZPE4DtpWsN6RtXbaJiDZgP1BPlkgOATuArcA3I6Ixx1j71c4NawAY44KIZlal8kwe6mJb9LDNbOAYMBGYAnxR0nEFoiTdKGmVpFW7d+/ua7z95tCrrwAw8dz3VTgSM7N85Jk8GoBJJeunA7/urk2aohoFNAJ/DPwkIo5GxC7geWBW5zeIiAciYlZEzBo/fnwOXeidY5vf4HCtGDHxjEqHYmaWizyTx0pgmqQpkgYD1wOLO7VZDMxPy9cCz0ZEkE1VXazMMGAO8EqOsfarIdt2s++dw8gO35iZVZ/ckkc6hnEzsBTYAPwoItZLukvSFanZQ0C9pE3AAqDjdN77gOHAOrIk9N2IWJtXrP1tzI5DtEx6++wJmZn1t5o8XzwilgBLOm37aslyC9lpuZ2fd7Cr7aeCA7u3M/JQO3uneMrKzKqXrzDvZ9vXvQDA8GnvrnAkZmb5cfLoZ40bstm18dNnVjgSM7P8OHn0s5ZNG2kZBJOmzqh0KGZmuXHy6Gd6Yzs7x9cwfMiISodiZpabwiePaG1ly6f/hL0PPUzb3r19fr1h2xs5MNH3LDez6lb45NG2Zw9x7Bi7vvENNn54Lg233ELzihVEW1uPX6N1yxb2fOc7bL7uDxm5/yhtZ74rx4jNzCov11N1TwWDJk5k8mM/4Mhrr9G0cBH7n3qK5qd/Ss2ECYy66ipGX3M1g884/rTbI5s2cWDZMpqXPc2RV7LrFw+c/Q7+be4Azr5i7knuhZnZyaXsgu5T36xZs2LVqlV9fp1obaV5xQqaFi3i0HM/h/Z2hs6ezehrr2HwWWfT/MxPaV72NK2vvQYStTMu4PULJnDfqNVsrG3isjMv486L7mTUEE9dmdnbn6TVEXFc+acTPs/Jo3tHd+5k/49/TNOiJzi6LRUIHjCAoRdeyIjLLmPtOXXcs+V7bN6/mZkTZrJg1gLOH39+v8ZgZpYnJ48ckkeHaG/n8C9X0rbzNwz74Ad5+ZQmd1MAAAgUSURBVFgD31r1LdbsWsPkkZO57b238ZFJH3EtKzM75fQ2eRT+mEdLWwt3PHcHY2vHMrZubPY9fdXX1VNfW8+IwSMYNud9bD2wlbvWfI1lW5ZRX1vPV+Z8haunXU3NgML/GM2sYAr/qXfw6EE279/M6p2raTrSRBx3yxGoUQ1jasewr2UfgwYO4qbzb+KGc25g6KChFYjYzKzyCp88xtWN48krnwSgrb2NpiNNNLY00tjSyN7f7v2/5caWRkYMGsH8c+Yzfqgr5ppZsRU+eZSqGVDDuLpxjKsbV+lQzMze1gp/kaCZmZXPycPMzMrm5GFmZmVz8jAzs7LlmjwkfUzSq5I2SfpyF48PkfTD9PgLkiaXPHaepP+WtF7SS5Jq84zVzMx6LrfkIWkgcB9wOTAd+KSk6Z2afRbYFxFTgXuBu9Nza4BHgc9FxDnAXOBoXrGamVl58tzzmA1siojXI6IVeByY16nNPOCRtLwQuERZjY+PAmsj4lcAEbE3Io7lGKuZmZUhz+RxGrCtZL0hbeuyTUS0AfuBeuB3gJC0VNIaSbfnGKeZmZUpz4sEu6oS2Ln2R3dtaoAPABcCh4FnUvGuZ97yZOlG4Ma0elDSq32Idxywpw/PP5W578VV5P4Xue/wZv/P7M2T80weDcCkkvXTgV9306YhHecYBTSm7f8VEXsAJC0BZgJvSR4R8QDwQH8EK2lVbypLVgP3vZh9h2L3v8h9h773P89pq5XANElTJA0GrgcWd2qzGJiflq8Fno2sRvxS4DxJQ1NS+TDwco6xmplZGXLb84iINkk3kyWCgcDDEbFe0l3AqohYDDwE/KukTWR7HNen5+6TdA9ZAgpgSUT8R16xmplZeXItjBgRS4AlnbZ9tWS5Bbium+c+Sna67snSL9Nfpyj3vbiK3P8i9x362P+quZOgmZmdPC5PYmZmZSt88jhRCZVqI2mSpOWSNqTSL19I28dKelrSxvR9TKVjzYukgZJelPTvaX1KKo+zMZXLGVzpGPMgabSkhZJeSeN/UcHG/bb0O79O0mOSaqt17CU9LGmXpHUl27oca2W+nT4D10qa2ZP3KHTy6GEJlWrTBnwxIt4DzAH+MvX5y8AzETGN7JToak6kXwA2lKzfDdyb+r6PrGxONfoH4CcR8W7gfLKfQSHGXdJpwOeBWRFxLtlJPNdTvWP/PeBjnbZ1N9aXA9PS143A/T15g0InD3pWQqWqRMSOiFiTlpvJPkBO462lYh4BrqxMhPmSdDrwB8CDaV3AxWTlcaBK+y5pJPAhsjMciYjWiGiiIOOe1AB16fT/ocAOqnTsI+JnZGewlupurOcB/xKZ/wFGS3rXid6j6MmjJyVUqlaqYjwDeAF4R0TsgCzBABMqF1mu/h64HWhP6/VAUyqPA9X7O3AWsBv4bpqye1DSMAoy7hGxHfgmsJUsaewHVlOMse/Q3Vj36nOw6MmjJyVUqpKk4cAi4NaIOFDpeE4GSZ8AdkXE6tLNXTStxt+BGrIqDfdHxAzgEFU6RdWVNL8/D5gCTASGkU3XdFaNY38ivfobKHry6EkJlaojaRBZ4vh+RDyRNu/s2FVN33dVKr4cvR+4QtIbZFOUF5PtiYxOUxlQvb8DDUBDRLyQ1heSJZMijDvApcDmiNgdEUeBJ4Dfpxhj36G7se7V52DRk0dPSqhUlTTH/xCwISLuKXmotFTMfOCpkx1b3iLijog4PSImk431sxHxKWA5WXkcqN6+/wbYJul306ZLyEr+VP24J1uBOankkXiz/1U/9iW6G+vFwJ+ms67mAPs7prf+P4W/SFDSx8n+++woofK1CoeUK0kfAJ4DXuLNef+/Ijvu8SPgDLI/tOsiovMBt6ohaS7wpYj4hKSzyPZExgIvAp+OiCOVjC8Pki4gO1FgMPA68BmyfyALMe6S/hr4I7IzDl8E/pxsbr/qxl7SY2Q30RsH7ATuBJ6ki7FOyfSfyM7OOgx8JiJWnfA9ip48zMysfEWftjIzs15w8jAzs7I5eZiZWdmcPMzMrGxOHmZmVjYnDzNA0sEcXnOFpOPuES3pVklDS9aXSBrd3+9vlicnD7OT71aywnwARMTHU5FCs1OGk4dZNySNl7RI0sr09f60fbakX6QCg7/ouGpbUp2kx9M9EX4I1HXxmp8nq620XNLytO0NSeMkTU732ngw3XPi+5IulfR8ugfD7NR+WLpfw8oUQ1VXgra3J18kaEY2bRURwztt+wHwzxHxc0lnAEsj4j2pvPnhiGiTdClwU0RcI2kBcG5E/Jmk84A1wJzOV+um2lqzImJP6TowHNhEVul4PVn5nF+R3WPiCrIrf6+U9LfAyxHxaJru+iUwIyIO5fLDMetCzYmbmBXWpcD0rHoDACMljQBGAY9ImkZWfXRQevxDwLcBImKtpLW9eM/NEfESgKT1ZDfvCUkvAZNTm4+SFXj8UlqvJSs5saHzi5nlxcnDrHsDgIsi4relGyX9I7A8Iq5K90RZUfJwX3flS+sqtZest/Pm36uAayLi1T6+l1mv+ZiHWfeWATd3rKTCgpDteWxPyzeUtP8Z8KnU9lzgvG5etxkY0Ye4lgK3pIJ2SJrRh9cy6xUnD7PMUEkNJV8LSPe8TgfAXwY+l9p+Hfg7Sc+TVWPucD8wPE1X3U52LKIrDwD/2XHAvBf+hmyqbK2kdWnd7KTyAXMzMyub9zzMzKxsTh5mZlY2Jw8zMyubk4eZmZXNycPMzMrm5GFmZmVz8jAzs7I5eZiZWdn+F2/fo+gaABucAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_rmse_lead(y_train, lr.predict(X_train), 'lr train')\n",
    "plot_rmse_lead(y_valid, lr.predict(X_valid), 'lr valid')\n",
    "plot_rmse_lead(y_train, lr_obs.predict(Xconcat_train), 'lr_obs train')\n",
    "plot_rmse_lead(y_valid, lr_obs.predict(Xconcat_valid), 'lr_obs valid')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "These results seam odd to me. Did we accidentally cheat by including observations we do not have? Let's ask Martin?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Your turn\n",
    "\n",
    "- Try to get the best score possible. As you will have noticed, I am not really able to give you many tips, so be creative and try a lot of things out!\n",
    "- Use the observation data for a neural network.\n",
    "- You could for example play around with early stopping, ensembles and dropout. Or combine different types of models in an ensemble."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  },
  "toc": {
   "nav_menu": {},
   "number_sections": true,
   "sideBar": false,
   "skip_h1_title": true,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}