texas_thunder_augmented.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h1>Thunderstorms in Texas with income and storm order</h1>\n",
    "\n",
    "<p>In this notebook, we add both mean income per county and \"storm order\" as input features to the data. The idea is that we would like to found out how much (if any) these features can improve our data.</p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# packages used\n",
    "import pandas as pd\n",
    "import torch\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('seaborn-whitegrid')\n",
    "%matplotlib inline\n",
    "import pylab\n",
    "pylab.rcParams['figure.figsize'] = (8.25, 6)\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 352 entries, 0 to 351\n",
      "Data columns (total 13 columns):\n",
      "CZ_NAME_STR            352 non-null object\n",
      "BEGIN_LOCATION         352 non-null object\n",
      "BEGIN_DATE             352 non-null object\n",
      "BEGIN_TIME             352 non-null int64\n",
      "MAGNITUDE              352 non-null float64\n",
      "STORM_ORDER            352 non-null int64\n",
      "DAMAGE_PROPERTY_NUM    352 non-null int64\n",
      "DAMAGE_CROPS_NUM       352 non-null int64\n",
      "END_LOCATION           352 non-null object\n",
      "BEGIN_LAT              352 non-null float64\n",
      "BEGIN_LON              352 non-null float64\n",
      "END_LAT                352 non-null float64\n",
      "END_LON                352 non-null float64\n",
      "dtypes: float64(5), int64(4), object(4)\n",
      "memory usage: 35.8+ KB\n"
     ]
    }
   ],
   "source": [
    "# load top 500 thunderstorms in texas from 1950 - 2018\n",
    "data = pd.read_csv(\"texas_thunderstorm.csv\", usecols=[1, 2, 3, 4, 6, 7, 10, 11, 29, 30, 31, 32, 33])\n",
    "data.info()\n",
    "\n",
    "# Randomize the data\n",
    "data = data.sample(frac=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "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>DAMAGE_PROPERTY_NUM</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>301</th>\n",
       "      <td>3000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>291</th>\n",
       "      <td>250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>5000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>326</th>\n",
       "      <td>1000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>139</th>\n",
       "      <td>5000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>310</th>\n",
       "      <td>800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>63</th>\n",
       "      <td>1500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>158</th>\n",
       "      <td>350000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>166</th>\n",
       "      <td>250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>208</th>\n",
       "      <td>1000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>106</th>\n",
       "      <td>5000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>332</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>334</th>\n",
       "      <td>1500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>400000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>84</th>\n",
       "      <td>200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>172</th>\n",
       "      <td>4000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>286</th>\n",
       "      <td>200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>252</th>\n",
       "      <td>189000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>56</th>\n",
       "      <td>2000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>257</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>110</th>\n",
       "      <td>350000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77</th>\n",
       "      <td>250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>295</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>174</th>\n",
       "      <td>200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>225</th>\n",
       "      <td>450000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>230</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>88</th>\n",
       "      <td>300000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>234</th>\n",
       "      <td>200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>278</th>\n",
       "      <td>15000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>339</th>\n",
       "      <td>750000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>182</th>\n",
       "      <td>300000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>242</th>\n",
       "      <td>700000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>198</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>206</th>\n",
       "      <td>350000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>260</th>\n",
       "      <td>200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>338</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>144</th>\n",
       "      <td>250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>224</th>\n",
       "      <td>250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>218</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>183</th>\n",
       "      <td>300000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>303</th>\n",
       "      <td>200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>137</th>\n",
       "      <td>200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>185</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>104</th>\n",
       "      <td>375000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>204</th>\n",
       "      <td>5000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>299</th>\n",
       "      <td>200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>216</th>\n",
       "      <td>500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>282</th>\n",
       "      <td>250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97</th>\n",
       "      <td>1100000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>330</th>\n",
       "      <td>1000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>352 rows × 1 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     DAMAGE_PROPERTY_NUM\n",
       "301              3000000\n",
       "291               250000\n",
       "18               5000000\n",
       "326              1000000\n",
       "139              5000000\n",
       "310               800000\n",
       "63               1500000\n",
       "158               350000\n",
       "166               250000\n",
       "208              1000000\n",
       "106              5000000\n",
       "332               500000\n",
       "334              1500000\n",
       "0                 400000\n",
       "84                200000\n",
       "172              4000000\n",
       "286               200000\n",
       "252               189000\n",
       "56               2000000\n",
       "257               500000\n",
       "115               500000\n",
       "110               350000\n",
       "77                250000\n",
       "295               500000\n",
       "174               200000\n",
       "225               450000\n",
       "35                500000\n",
       "230               500000\n",
       "88                300000\n",
       "234               200000\n",
       "..                   ...\n",
       "278             15000000\n",
       "339               750000\n",
       "182               300000\n",
       "242               700000\n",
       "198               500000\n",
       "19                200000\n",
       "206               350000\n",
       "260               200000\n",
       "338               500000\n",
       "25                500000\n",
       "101               500000\n",
       "144               250000\n",
       "75                500000\n",
       "224               250000\n",
       "218               500000\n",
       "183               300000\n",
       "14                500000\n",
       "303               200000\n",
       "72                500000\n",
       "137               200000\n",
       "185               500000\n",
       "104               375000\n",
       "36                200000\n",
       "204              5000000\n",
       "299               200000\n",
       "216               500000\n",
       "6                 250000\n",
       "282               250000\n",
       "97               1100000\n",
       "330              1000000\n",
       "\n",
       "[352 rows x 1 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# The outcomes = what we're trying to predict\n",
    "thunderstorm_outcomes = data.iloc[:, [6]]\n",
    "\n",
    "list(thunderstorm_outcomes)\n",
    "thunderstorm_outcomes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['BEGIN_TIME',\n",
       " 'MAGNITUDE',\n",
       " 'STORM_ORDER',\n",
       " 'DAMAGE_CROPS_NUM',\n",
       " 'BEGIN_LAT',\n",
       " 'BEGIN_LON',\n",
       " 'END_LAT',\n",
       " 'END_LON']"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# The predictors = the features we are using to predict property damage in Texas\n",
    "thunderstorm_predictors = data.drop(['DAMAGE_PROPERTY_NUM',\n",
    "                                     'BEGIN_LOCATION',\n",
    "                                     'END_LOCATION',\n",
    "                                     'CZ_NAME_STR',\n",
    "                                     'BEGIN_DATE'], axis=1)\n",
    "\n",
    "list(thunderstorm_predictors)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 3302 entries, 0 to 3301\n",
      "Data columns (total 2 columns):\n",
      "name        3302 non-null object\n",
      "estimate    3301 non-null float64\n",
      "dtypes: float64(1), object(1)\n",
      "memory usage: 51.7+ KB\n"
     ]
    }
   ],
   "source": [
    "# load mean income data per county\n",
    "texas_income_data = pd.read_csv(\"MACHINE_LEARNING_DATA/county_year/texas_median_county_income_all.csv\")\n",
    "texas_income_data.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Add column to input features\n",
    "# 1990, 2000, 2009 forward, using the closest year available if the exact year does not exist\n",
    "def parse_date(year):\n",
    "    if int(year) < 95:\n",
    "        return '1990'\n",
    "    elif int(year) >= 95 or int(year) < 5:\n",
    "        return '2000'\n",
    "    elif int(year) >= 5 and int(year) < 9:\n",
    "        return '2009'\n",
    "    elif int(year) >= 9:\n",
    "        return '20' + year\n",
    "    \n",
    "# add a mean income column to the input features\n",
    "def add_mean_income(thunderstorm_predictors):\n",
    "    thunderstorm_predictors['MEAN_INCOME'] = 100\n",
    "    years = [date.split('/')[-1] for date in data['BEGIN_DATE']]\n",
    "    years = [parse_date(year) for year in years]\n",
    "    estimates = []\n",
    "    for i in range(len(thunderstorm_predictors)):\n",
    "        keyword = data['CZ_NAME_STR'][i] + years[i]\n",
    "        estimate = texas_income_data[texas_income_data['name'] == keyword].values[0][1]\n",
    "        estimates.append(estimate)\n",
    "    thunderstorm_predictors['MEAN_INCOME'] = estimates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "add_mean_income(thunderstorm_predictors)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1a1a698ac8>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "corr = thunderstorm_predictors.corr()\n",
    "\n",
    "# plot correlation matrix\n",
    "f, ax = plt.subplots(figsize=(10, 8))\n",
    "sns.heatmap(corr, mask=np.zeros_like(corr, dtype=np.bool), cmap=sns.diverging_palette(220, 10, as_cmap=True),\n",
    "            square=True, ax=ax)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 9)\n",
      "Shape of inputs array test (71, 9)\n"
     ]
    }
   ],
   "source": [
    "def preprocess(thunderstorm_outcomes, thunderstorm_predictors):\n",
    "    '''\n",
    "        This function takes in both the input features and the outcome data for the property damage.\n",
    "        The input features are all normalized by subtracting the mean and dividing by the standard \n",
    "        deviation. The outcome data is normalized by first transforming it to a log scale and then doing\n",
    "        mean normalization. \n",
    "    '''\n",
    "    # Make the outcomes into a numpy array\n",
    "    outcome_array = thunderstorm_outcomes.values.astype(np.float)\n",
    "    \n",
    "    # log scale\n",
    "    outcome_array = np.log(outcome_array + 1)\n",
    "\n",
    "    # split into test and training data\n",
    "    split_index = int(0.8 * len(outcome_array))\n",
    "    outcome_array_train = outcome_array[:split_index, :]\n",
    "    outcome_array_test = outcome_array[split_index:, :]\n",
    "    print(outcome_array_train.shape)\n",
    "    print('Shape of outcome array training', outcome_array_train.shape)\n",
    "    print('Shape of outcome array test', outcome_array_test.shape)\n",
    "    \n",
    "    # normalize outcome\n",
    "    outcome_array_train = (outcome_array_train - np.mean(outcome_array))/np.std(outcome_array)\n",
    "    outcome_array_test = (outcome_array_test - np.mean(outcome_array))/np.std(outcome_array)\n",
    "    \n",
    "    outcome_mean_log = np.mean(outcome_array)\n",
    "    outcome_std_log = np.std(outcome_array)\n",
    "\n",
    "    # Make that numpy array into a torch Tensor\n",
    "    outcome_Tensor_train = torch.from_numpy(outcome_array_train)\n",
    "    outcome_Tensor_test = torch.from_numpy(outcome_array_test)\n",
    "\n",
    "    # Make the predictors into a numpy array\n",
    "    predictors_array = thunderstorm_predictors.values\n",
    "    # NOTE: we took out any missing data values in the csv \n",
    "\n",
    "    # split into test and training data\n",
    "    predictors_array_train = predictors_array[:split_index, :]\n",
    "    predictors_array_test = predictors_array[split_index:, :]\n",
    "    print('Shape of inputs array training', predictors_array_train.shape)\n",
    "    print('Shape of inputs array test', predictors_array_test.shape)\n",
    "\n",
    "    # Normalize inputs\n",
    "    predictors_array_train = (predictors_array_train - np.mean(predictors_array, axis=0))/np.std(predictors_array, axis=0)\n",
    "    predictors_array_test = (predictors_array_test - np.mean(predictors_array, axis=0))/np.std(predictors_array, axis=0)\n",
    "\n",
    "    # Make that numpy array into a torch Tensor\n",
    "    predictors_Tensor_train = torch.from_numpy(predictors_array_train.astype(np.float))\n",
    "    predictors_Tensor_test = torch.from_numpy(predictors_array_test.astype(np.float))\n",
    "\n",
    "    # Convert those DoubleTensors to FloatTensors\n",
    "    # Chaning outcomes from a DoubleTensor to a FloatTensor\n",
    "    outcome_Tensor_train = outcome_Tensor_train.float()\n",
    "    outcome_Tensor_test = outcome_Tensor_test.float()\n",
    "\n",
    "    # Changing the predictors too\n",
    "    predictors_Tensor_train = predictors_Tensor_train.float()\n",
    "    predictors_Tensor_test = predictors_Tensor_test.float()\n",
    "    \n",
    "    return (predictors_Tensor_train, predictors_Tensor_test, outcome_Tensor_train,\n",
    "            outcome_Tensor_test, outcome_mean_log, outcome_std_log)\n",
    "\n",
    "(predictors_Tensor_train, predictors_Tensor_test, outcome_Tensor_train,\\\n",
    "outcome_Tensor_test, outcome_mean_log, outcome_std_log) = preprocess(thunderstorm_outcomes, thunderstorm_predictors)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h1>One hidden layer</h1>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "class one_hidden_Net(torch.nn.Module):\n",
    "    def __init__(self):\n",
    "        super(one_hidden_Net, self).__init__()\n",
    "        self.fc1 = torch.nn.Linear(9, 6)\n",
    "        self.fc2 = torch.nn.Linear(6, 1)\n",
    "        \n",
    "    def forward(self, x):\n",
    "        x = torch.nn.functional.relu(self.fc1(x))\n",
    "        x = self.fc2(x)\n",
    "        return(x)\n",
    "    \n",
    "# Make it\n",
    "one_hidden_model = one_hidden_Net()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.7/site-packages/torch/nn/_reduction.py:49: UserWarning: size_average and reduce args will be deprecated, please use reduction='sum' instead.\n",
      "  warnings.warn(warning.format(ret))\n"
     ]
    }
   ],
   "source": [
    "# Define the loss function\n",
    "loss_fn = torch.nn.MSELoss(size_average = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_neural_net(model, train_outcomes, train_inputs, test_outcomes, test_inputs):\n",
    "    # Make an empty list to add to for each iteration\n",
    "    current_loss_value_mini = []\n",
    "    \n",
    "    # For the cv too\n",
    "    current_test_loss_value = []\n",
    "    \n",
    "    # Set the optimizer\n",
    "    # originally step size was 1e-2\n",
    "    optimizer = torch.optim.Adagrad(model.parameters(), lr = 1e-2)\n",
    "    \n",
    "    # Train the neural network\n",
    "    # 100000 iterations seems to give much better results when networks is two layers deep\n",
    "    for t in range(20000):\n",
    "        # Get the current mini-batch indices\n",
    "        indices = torch.randperm(len(train_inputs))[0:50]\n",
    "        \n",
    "        # Get the current mini-batches\n",
    "        predictors_mini = train_inputs[indices]\n",
    "        outcome_mini = train_outcomes[indices]\n",
    "        \n",
    "        # Get the current predictions\n",
    "        current_pred_mini = model(predictors_mini)\n",
    "        test_pred = model(test_inputs)\n",
    "        \n",
    "        # Compute current losses and and append that to the empty list\n",
    "        current_loss_mini = loss_fn(current_pred_mini, outcome_mini)\n",
    "        current_loss_value_mini.append(current_loss_mini.data.item())\n",
    "        \n",
    "        current_test_loss = loss_fn(test_pred, test_outcomes)\n",
    "        current_test_loss_value.append(current_test_loss.data.item())\n",
    "        \n",
    "        # Zero the gradients before backwards pass\n",
    "        optimizer.zero_grad()\n",
    "        \n",
    "        # Backwards pass\n",
    "        current_loss_mini.backward()\n",
    "        \n",
    "        # Update model parameters\n",
    "        optimizer.step()\n",
    "        \n",
    "    # Get iterations as a list for x-axis of plot\n",
    "    iterations = list(range(0, 20000))\n",
    "    \n",
    "    # Convert total square error to mean square error\n",
    "    mse = (np.asarray(current_loss_value_mini) / len(predictors_mini)).tolist()\n",
    "    \n",
    "    # For cv too\n",
    "    mse_test = (np.asarray(current_test_loss_value) / len(test_inputs)).tolist()\n",
    "    \n",
    "    # Plot loss versus iteration\n",
    "    plt.plot(iterations, mse)\n",
    "    plt.plot(iterations, mse_test, \"r-\")\n",
    "    plt.xlabel(\"Iteration\", size = 14)\n",
    "    plt.ylabel(\"Loss Value\", size = 14)\n",
    "    plt.xticks(fontsize = 12)\n",
    "    plt.yticks(fontsize = 12)\n",
    "    plt.title(\"How'd the Optimizing Go?\", size = 16)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_neural_net(one_hidden_model, outcome_Tensor_train, predictors_Tensor_train, \n",
    "                 outcome_Tensor_test, predictors_Tensor_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_train_perf(train_predictors, train_obs, model):\n",
    "    \n",
    "    # Plots how well the model performs on training data\n",
    "    # train_predictors = the values that the model predicts from, PyTorch Tensor\n",
    "    # train_obs = what the model would ideally predict, PyTorch Tensor\n",
    "    # Returns a plot\n",
    "    \n",
    "    \n",
    "    # Get the predictions\n",
    "    train_predictors.requires_grad = False\n",
    "    train_predictions = model(train_predictors)\n",
    "    \n",
    "    # Convert that to a numpy array\n",
    "    train_predictions = train_predictions.data.numpy()\n",
    "    \n",
    "    # Put it in log-transformed scale and convert it to a list\n",
    "    train_predictions = (train_predictions*outcome_std_log + outcome_mean_log).tolist()\n",
    "    \n",
    "    \n",
    "    # Put the observed outcomes in log-transformed scale\n",
    "    # and convert it to a list\n",
    "    train_obs = (train_obs.numpy()*outcome_std_log + outcome_mean_log).tolist()\n",
    "    \n",
    "    \n",
    "    # Plot it\n",
    "    plt.scatter(train_obs, train_predictions, facecolors = \"none\", edgecolors = \"k\")\n",
    "    axes = plt.gca()\n",
    "    axes.set_xlim([-1, 24])\n",
    "    axes.set_ylim([-1, 24])\n",
    "    axes.set_yticks([0, 3, 6, 9, 12, 15, 18, 21, 24])\n",
    "    plt.xticks(fontsize = 12)\n",
    "    plt.yticks(fontsize = 12)\n",
    "    plt.plot([-1, 24], [-1, 24], 'k-', lw = 2)\n",
    "    axes.set_xlabel(\"Observed (Log-Transformed US Dollars)\", size = 14)\n",
    "    axes.set_ylabel(\"Predicted (Log-Transformed US Dollars)\", size = 14)\n",
    "    axes.set_title(\"Model Performance on Training Data\", size = 16)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "ename": "RuntimeError",
     "evalue": "size mismatch, m1: [281 x 8], m2: [9 x 6] at /Users/soumith/mc3build/conda-bld/pytorch_1549597882250/work/aten/src/TH/generic/THTensorMath.cpp:940",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mRuntimeError\u001b[0m                              Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-32-75d9e0175f28>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mplot_train_perf\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpredictors_Tensor_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moutcome_Tensor_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mone_hidden_model\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m<ipython-input-16-009a303463b1>\u001b[0m in \u001b[0;36mplot_train_perf\u001b[0;34m(train_predictors, train_obs, model)\u001b[0m\n\u001b[1;32m      9\u001b[0m     \u001b[0;31m# Get the predictions\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     10\u001b[0m     \u001b[0mtrain_predictors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrequires_grad\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mFalse\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 11\u001b[0;31m     \u001b[0mtrain_predictions\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrain_predictors\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     12\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     13\u001b[0m     \u001b[0;31m# Convert that to a numpy array\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda3/lib/python3.7/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    487\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    488\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 489\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    490\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    491\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-12-0bf70aa6a379>\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, x)\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 8\u001b[0;31m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfunctional\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrelu\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfc1\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      9\u001b[0m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfc2\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     10\u001b[0m         \u001b[0;32mreturn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda3/lib/python3.7/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    487\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    488\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 489\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    490\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    491\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda3/lib/python3.7/site-packages/torch/nn/modules/linear.py\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m     65\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0mweak_script_method\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     66\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 67\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mF\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlinear\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mweight\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbias\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     68\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     69\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mextra_repr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py\u001b[0m in \u001b[0;36mlinear\u001b[0;34m(input, weight, bias)\u001b[0m\n\u001b[1;32m   1350\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdim\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m2\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mbias\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1351\u001b[0m         \u001b[0;31m# fused op is marginally faster\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1352\u001b[0;31m         \u001b[0mret\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0maddmm\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mjit\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_unwrap_optional\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbias\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mweight\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mt\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1353\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1354\u001b[0m         \u001b[0moutput\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmatmul\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mweight\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mt\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mRuntimeError\u001b[0m: size mismatch, m1: [281 x 8], m2: [9 x 6] at /Users/soumith/mc3build/conda-bld/pytorch_1549597882250/work/aten/src/TH/generic/THTensorMath.cpp:940"
     ]
    }
   ],
   "source": [
    "plot_train_perf(predictors_Tensor_train, outcome_Tensor_train, one_hidden_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "def compute_MSE(predictors, observed_outcomes, model):\n",
    "    \n",
    "    # Computes the mean squared error of the model\n",
    "    # predictors = the variables and values used in prediction\n",
    "    # observed_outcomes = what the prediction would ideally yield\n",
    "    # Returns a PyTorch Variable\n",
    "    \n",
    "    \n",
    "    # Get the validation set predictions\n",
    "    predictors.requires_grad = False\n",
    "    predictions = model(predictors)\n",
    "    \n",
    "    # Get the validation set outcomes\n",
    "    outcomes = observed_outcomes\n",
    "    \n",
    "    # Get the loss - (total) residual sum of squares\n",
    "    total_SE = loss_fn(predictions, outcomes)\n",
    "    \n",
    "    # Average that to get mean square error\n",
    "    mean_SE = total_SE / len(predictors)\n",
    "    return(mean_SE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "ename": "RuntimeError",
     "evalue": "size mismatch, m1: [71 x 8], m2: [9 x 6] at /Users/soumith/mc3build/conda-bld/pytorch_1549597882250/work/aten/src/TH/generic/THTensorMath.cpp:940",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mRuntimeError\u001b[0m                              Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-34-7ac8f2bba544>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mone_hidden_MSE\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompute_MSE\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpredictors_Tensor_test\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moutcome_Tensor_test\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mone_hidden_model\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Mean squared error:\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mone_hidden_MSE\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-33-b065363eae23>\u001b[0m in \u001b[0;36mcompute_MSE\u001b[0;34m(predictors, observed_outcomes, model)\u001b[0m\n\u001b[1;32m      9\u001b[0m     \u001b[0;31m# Get the validation set predictions\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     10\u001b[0m     \u001b[0mpredictors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrequires_grad\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mFalse\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 11\u001b[0;31m     \u001b[0mpredictions\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpredictors\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     12\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     13\u001b[0m     \u001b[0;31m# Get the validation set outcomes\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda3/lib/python3.7/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    487\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    488\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 489\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    490\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    491\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-12-0bf70aa6a379>\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, x)\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 8\u001b[0;31m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfunctional\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrelu\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfc1\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      9\u001b[0m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfc2\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     10\u001b[0m         \u001b[0;32mreturn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda3/lib/python3.7/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    487\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    488\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 489\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    490\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    491\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda3/lib/python3.7/site-packages/torch/nn/modules/linear.py\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m     65\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0mweak_script_method\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     66\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 67\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mF\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlinear\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mweight\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbias\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     68\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     69\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mextra_repr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py\u001b[0m in \u001b[0;36mlinear\u001b[0;34m(input, weight, bias)\u001b[0m\n\u001b[1;32m   1350\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdim\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m2\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mbias\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1351\u001b[0m         \u001b[0;31m# fused op is marginally faster\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1352\u001b[0;31m         \u001b[0mret\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0maddmm\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mjit\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_unwrap_optional\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbias\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mweight\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mt\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1353\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1354\u001b[0m         \u001b[0moutput\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmatmul\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mweight\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mt\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mRuntimeError\u001b[0m: size mismatch, m1: [71 x 8], m2: [9 x 6] at /Users/soumith/mc3build/conda-bld/pytorch_1549597882250/work/aten/src/TH/generic/THTensorMath.cpp:940"
     ]
    }
   ],
   "source": [
    "one_hidden_MSE = compute_MSE(predictors_Tensor_test, outcome_Tensor_test, one_hidden_model)\n",
    "print(\"Mean squared error:\", one_hidden_MSE.item())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "def display_readable_predictions(predictors, observed_outcomes, model):\n",
    "    '''\n",
    "        This function gives us a better idea of how accurate our predictions are on the \"real\" (non-log)\n",
    "        scale.\n",
    "    '''\n",
    "    predictions = np.exp((model(predictors) * outcome_std_log + outcome_mean_log).detach().numpy()) - 1\n",
    "    \n",
    "    observed = np.exp(observed_outcomes * outcome_std_log + outcome_mean_log) - 1\n",
    "    \n",
    "    for i in range(len(predictions)):\n",
    "        print((predictions[i][0].item(), observed[i][0].item()))\n",
    "        print(predictions[i][0].item() - observed[i][0].item())\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(601773.0, 199999.984375)\n",
      "401773.015625\n",
      "(6632110.0, 210000.0)\n",
      "6422110.0\n",
      "(1801991.75, 20000004.0)\n",
      "-18198012.25\n",
      "(417850.1875, 250000.0)\n",
      "167850.1875\n",
      "(507413.75, 500000.28125)\n",
      "7413.46875\n",
      "(554017.125, 250000.0)\n",
      "304017.125\n",
      "(410441.6875, 220000.09375)\n",
      "190441.59375\n",
      "(892159.8125, 1500000.75)\n",
      "-607840.9375\n",
      "(521007.5625, 2000000.375)\n",
      "-1478992.8125\n",
      "(587730.3125, 199999.984375)\n",
      "387730.328125\n",
      "(557998.75, 325000.03125)\n",
      "232998.71875\n",
      "(648389.1875, 300000.1875)\n",
      "348389.0\n",
      "(308564.21875, 199999.984375)\n",
      "108564.234375\n",
      "(416224.71875, 250000.0)\n",
      "166224.71875\n",
      "(425251.65625, 5000005.0)\n",
      "-4574753.34375\n",
      "(1193779.5, 300000.1875)\n",
      "893779.3125\n",
      "(448805.90625, 250000.0)\n",
      "198805.90625\n",
      "(994282.125, 1500000.75)\n",
      "-505718.625\n",
      "(1017276.625, 5000005.0)\n",
      "-3982728.375\n",
      "(363149.375, 500000.28125)\n",
      "-136850.90625\n",
      "(646720.5625, 250000.0)\n",
      "396720.5625\n",
      "(375092.78125, 2000000.375)\n",
      "-1624907.59375\n",
      "(556517.625, 250000.0)\n",
      "306517.625\n",
      "(403699.75, 1500000.75)\n",
      "-1096301.0\n",
      "(354990.8125, 275000.09375)\n",
      "79990.71875\n",
      "(673456.625, 500000.28125)\n",
      "173456.34375\n",
      "(631786.875, 1000000.125)\n",
      "-368213.25\n",
      "(867366.25, 1000000.125)\n",
      "-132633.875\n",
      "(196479.078125, 1000000.125)\n",
      "-803521.046875\n",
      "(336045.625, 1000000.125)\n",
      "-663954.5\n",
      "(351380.40625, 500000.28125)\n",
      "-148619.875\n",
      "(482197.46875, 500000.28125)\n",
      "-17802.8125\n",
      "(501017.46875, 250000.0)\n",
      "251017.46875\n",
      "(399361.9375, 300000.1875)\n",
      "99361.75\n",
      "(466043.0, 9999996.0)\n",
      "-9533953.0\n",
      "(716783.5625, 199999.984375)\n",
      "516783.578125\n",
      "(356302.71875, 800000.6875)\n",
      "-443697.96875\n",
      "(426951.875, 199999.984375)\n",
      "226951.890625\n",
      "(400174.0, 3000001.0)\n",
      "-2599827.0\n",
      "(1376407.125, 9999996.0)\n",
      "-8623588.875\n",
      "(315044.375, 199999.984375)\n",
      "115044.390625\n",
      "(2339549.0, 15000010.0)\n",
      "-12660461.0\n",
      "(497950.28125, 750000.25)\n",
      "-252049.96875\n",
      "(561953.3125, 300000.1875)\n",
      "261953.125\n",
      "(1701277.25, 700000.3125)\n",
      "1001276.9375\n",
      "(362023.8125, 500000.28125)\n",
      "-137976.46875\n",
      "(723973.0, 199999.984375)\n",
      "523973.015625\n",
      "(1610615.5, 350000.15625)\n",
      "1260615.34375\n",
      "(435146.9375, 199999.984375)\n",
      "235146.953125\n",
      "(379565.3125, 500000.28125)\n",
      "-120434.96875\n",
      "(637028.6875, 500000.28125)\n",
      "137028.40625\n",
      "(399294.90625, 500000.28125)\n",
      "-100705.375\n",
      "(388878.59375, 250000.0)\n",
      "138878.59375\n",
      "(548355.375, 500000.28125)\n",
      "48355.09375\n",
      "(637125.3125, 250000.0)\n",
      "387125.3125\n",
      "(702400.375, 500000.28125)\n",
      "202400.09375\n",
      "(692885.0, 300000.1875)\n",
      "392884.8125\n",
      "(475605.3125, 500000.28125)\n",
      "-24394.96875\n",
      "(2385820.0, 199999.984375)\n",
      "2185820.015625\n",
      "(433393.0, 500000.28125)\n",
      "-66607.28125\n",
      "(681313.875, 199999.984375)\n",
      "481313.890625\n",
      "(555630.9375, 500000.28125)\n",
      "55630.65625\n",
      "(368470.15625, 375000.15625)\n",
      "-6530.0\n",
      "(940245.8125, 199999.984375)\n",
      "740245.828125\n",
      "(289428.875, 5000005.0)\n",
      "-4710576.125\n",
      "(593260.3125, 199999.984375)\n",
      "393260.328125\n",
      "(360332.25, 500000.28125)\n",
      "-139668.03125\n",
      "(1066762.0, 250000.0)\n",
      "816762.0\n",
      "(569739.5625, 250000.0)\n",
      "319739.5625\n",
      "(914602.4375, 1100000.25)\n",
      "-185397.8125\n",
      "(354833.4375, 1000000.125)\n",
      "-645166.6875\n"
     ]
    }
   ],
   "source": [
    "# Display (predicted, observed)\n",
    "#          difference\n",
    "display_readable_predictions(predictors_Tensor_test, outcome_Tensor_test, one_hidden_model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2>Two hidden layers</h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "class two_hidden_Net(torch.nn.Module):\n",
    "    def __init__(self):\n",
    "        super(two_hidden_Net, self).__init__()\n",
    "        self.fc1 = torch.nn.Linear(9, 6)\n",
    "        self.fc2 = torch.nn.Linear(6, 4)\n",
    "        self.fc3 = torch.nn.Linear(4, 1)\n",
    "        \n",
    "    def forward(self, x):\n",
    "        x = torch.nn.functional.relu(self.fc1(x))\n",
    "        x = torch.nn.functional.relu(self.fc2(x))\n",
    "        x = self.fc3(x)\n",
    "        return(x)\n",
    "    \n",
    "# Make it\n",
    "two_hidden_model = two_hidden_Net()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_neural_net(two_hidden_model, outcome_Tensor_train, predictors_Tensor_train,\n",
    "                outcome_Tensor_test, predictors_Tensor_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_train_perf(predictors_Tensor_train, outcome_Tensor_train, two_hidden_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mean squared error: 1.1282517910003662\n"
     ]
    }
   ],
   "source": [
    "MSE = compute_MSE(predictors_Tensor_test, outcome_Tensor_test, two_hidden_model)\n",
    "print('Mean squared error:', MSE.item())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(787235.8125, 199999.984375)\n",
      "587235.828125\n",
      "(6293555.5, 210000.0)\n",
      "6083555.5\n",
      "(2608166.75, 20000004.0)\n",
      "-17391837.25\n",
      "(368119.25, 250000.0)\n",
      "118119.25\n",
      "(811468.1875, 500000.28125)\n",
      "311467.90625\n",
      "(479378.96875, 250000.0)\n",
      "229378.96875\n",
      "(361511.46875, 220000.09375)\n",
      "141511.375\n",
      "(206634.828125, 1500000.75)\n",
      "-1293365.921875\n",
      "(308633.09375, 2000000.375)\n",
      "-1691367.28125\n",
      "(886245.0, 199999.984375)\n",
      "686245.015625\n",
      "(759072.625, 325000.03125)\n",
      "434072.59375\n",
      "(121529.2109375, 300000.1875)\n",
      "-178470.9765625\n",
      "(1157332.125, 199999.984375)\n",
      "957332.140625\n",
      "(437628.15625, 250000.0)\n",
      "187628.15625\n",
      "(760342.6875, 5000005.0)\n",
      "-4239662.3125\n",
      "(785555.125, 300000.1875)\n",
      "485554.9375\n",
      "(731286.9375, 250000.0)\n",
      "481286.9375\n",
      "(1920555.125, 1500000.75)\n",
      "420554.375\n",
      "(637288.75, 5000005.0)\n",
      "-4362716.25\n",
      "(432214.1875, 500000.28125)\n",
      "-67786.09375\n",
      "(1175795.25, 250000.0)\n",
      "925795.25\n",
      "(668171.75, 2000000.375)\n",
      "-1331828.625\n",
      "(407431.75, 250000.0)\n",
      "157431.75\n",
      "(301774.96875, 1500000.75)\n",
      "-1198225.78125\n",
      "(1504761.875, 275000.09375)\n",
      "1229761.78125\n",
      "(441798.9375, 500000.28125)\n",
      "-58201.34375\n",
      "(522273.65625, 1000000.125)\n",
      "-477726.46875\n",
      "(5578132.0, 1000000.125)\n",
      "4578131.875\n",
      "(448693.34375, 1000000.125)\n",
      "-551306.78125\n",
      "(328773.03125, 1000000.125)\n",
      "-671227.09375\n",
      "(366726.0625, 500000.28125)\n",
      "-133274.21875\n",
      "(624316.125, 500000.28125)\n",
      "124315.84375\n",
      "(432146.1875, 250000.0)\n",
      "182146.1875\n",
      "(341652.5, 300000.1875)\n",
      "41652.3125\n",
      "(297620.15625, 9999996.0)\n",
      "-9702375.84375\n",
      "(654929.5, 199999.984375)\n",
      "454929.515625\n",
      "(316766.125, 800000.6875)\n",
      "-483234.5625\n",
      "(300522.5, 199999.984375)\n",
      "100522.515625\n",
      "(351351.9375, 3000001.0)\n",
      "-2648649.0625\n",
      "(2168974.5, 9999996.0)\n",
      "-7831021.5\n",
      "(382630.21875, 199999.984375)\n",
      "182630.234375\n",
      "(554883.8125, 15000010.0)\n",
      "-14445126.1875\n",
      "(338970.125, 750000.25)\n",
      "-411030.125\n",
      "(481015.25, 300000.1875)\n",
      "181015.0625\n",
      "(1174943.25, 700000.3125)\n",
      "474942.9375\n",
      "(217962.265625, 500000.28125)\n",
      "-282038.015625\n",
      "(538990.1875, 199999.984375)\n",
      "338990.203125\n",
      "(1112134.875, 350000.15625)\n",
      "762134.71875\n",
      "(343216.4375, 199999.984375)\n",
      "143216.453125\n",
      "(328909.4375, 500000.28125)\n",
      "-171090.84375\n",
      "(752380.0625, 500000.28125)\n",
      "252379.78125\n",
      "(308356.53125, 500000.28125)\n",
      "-191643.75\n",
      "(379828.9375, 250000.0)\n",
      "129828.9375\n",
      "(689537.0, 500000.28125)\n",
      "189536.71875\n",
      "(906872.4375, 250000.0)\n",
      "656872.4375\n",
      "(448154.9375, 500000.28125)\n",
      "-51845.34375\n",
      "(521180.5, 300000.1875)\n",
      "221180.3125\n",
      "(448693.34375, 500000.28125)\n",
      "-51306.9375\n",
      "(3683971.0, 199999.984375)\n",
      "3483971.015625\n",
      "(439408.4375, 500000.28125)\n",
      "-60591.84375\n",
      "(635757.25, 199999.984375)\n",
      "435757.265625\n",
      "(434548.9375, 500000.28125)\n",
      "-65451.34375\n",
      "(272966.5, 375000.15625)\n",
      "-102033.65625\n",
      "(3612513.25, 199999.984375)\n",
      "3412513.265625\n",
      "(448693.34375, 5000005.0)\n",
      "-4551311.65625\n",
      "(833963.75, 199999.984375)\n",
      "633963.765625\n",
      "(564975.4375, 500000.28125)\n",
      "64975.15625\n",
      "(642521.75, 250000.0)\n",
      "392521.75\n",
      "(408500.5, 250000.0)\n",
      "158500.5\n",
      "(388355.3125, 1100000.25)\n",
      "-711644.9375\n",
      "(415935.4375, 1000000.125)\n",
      "-584064.6875\n"
     ]
    }
   ],
   "source": [
    "# Display (predicted, observed)\n",
    "#          difference\n",
    "display_readable_predictions(predictors_Tensor_test, outcome_Tensor_test, two_hidden_model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2>Feature importance evaluation</h2>\n",
    "<p>In order to analyze feature importance evaluations, we defined a new neural net with one less input feature. For every feature, we train the one hidden layer network and observe the change in mean squared error.</p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "class leave_out_Net(torch.nn.Module):\n",
    "    def __init__(self):\n",
    "        super(leave_out_Net, self).__init__()\n",
    "        self.fc1 = torch.nn.Linear(8, 5)\n",
    "        self.fc2 = torch.nn.Linear(5, 1)\n",
    "        \n",
    "    def forward(self, x):\n",
    "        x = torch.nn.functional.relu(self.fc1(x))\n",
    "        x = self.fc2(x)\n",
    "        return(x)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>Leave out BEGIN_TIME</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 8)\n",
      "Shape of inputs array test (71, 8)\n"
     ]
    },
    {
     "ename": "NameError",
     "evalue": "name 'one_hidden_mse' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-31-9e61f8789c75>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m     16\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     17\u001b[0m \u001b[0mone_layer_mse\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompute_MSE\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpredictors_Tensor_test\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moutcome_Tensor_test\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mwithout_time_model\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 18\u001b[0;31m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Difference with original mse\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mone_layer_mse\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mone_hidden_mse\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m: name 'one_hidden_mse' is not defined"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "thunderstorm_predictors = data.drop(['DAMAGE_PROPERTY_NUM',\n",
    "                                     'BEGIN_LOCATION',\n",
    "                                     'END_LOCATION',\n",
    "                                     'CZ_NAME_STR',\n",
    "                                     'BEGIN_DATE',\n",
    "                                     'BEGIN_TIME'], axis=1)\n",
    "add_mean_income(thunderstorm_predictors)\n",
    "(predictors_Tensor_train, predictors_Tensor_test, outcome_Tensor_train,\\\n",
    "outcome_Tensor_test, outcome_mean, outcome_std) = preprocess(thunderstorm_outcomes, thunderstorm_predictors)\n",
    "\n",
    "without_time_model = leave_out_Net()\n",
    "\n",
    "# Train one layer\n",
    "train_neural_net(without_time_model, outcome_Tensor_train, predictors_Tensor_train,\n",
    "            outcome_Tensor_test, predictors_Tensor_test)\n",
    "\n",
    "one_layer_mse = compute_MSE(predictors_Tensor_test, outcome_Tensor_test, without_time_model).item()\n",
    "print(\"Difference with original mse\", one_layer_mse - one_hidden_mse)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>Leave out DAMAGE_CROPS_NUM</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 600,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 8)\n",
      "Shape of inputs array test (71, 8)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.9643511772155762"
      ]
     },
     "execution_count": 600,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 594x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "thunderstorm_predictors = data.drop(['DAMAGE_PROPERTY_NUM',\n",
    "                                     'BEGIN_LOCATION',\n",
    "                                     'END_LOCATION',\n",
    "                                     'CZ_NAME_STR',\n",
    "                                     'BEGIN_DATE',\n",
    "                                     'DAMAGE_CROPS_NUM'], axis=1)\n",
    "add_mean_income(thunderstorm_predictors)\n",
    "(predictors_Tensor_train, predictors_Tensor_test, outcome_Tensor_train,\\\n",
    "outcome_Tensor_test, outcome_mean, outcome_std) = preprocess(thunderstorm_outcomes, thunderstorm_predictors)\n",
    "\n",
    "without_damage_crops_model = leave_out_Net()\n",
    "\n",
    "# Train one layer\n",
    "train_neural_net(without_damage_crops_model, outcome_Tensor_train, predictors_Tensor_train,\n",
    "            outcome_Tensor_test, predictors_Tensor_test)\n",
    "\n",
    "one_layer_mse = compute_MSE(predictors_Tensor_test, outcome_Tensor_test, without_damage_crops_model).item()\n",
    "print(\"Difference with original mse\", one_layer_mse - one_hidden_mse)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>Leave out STORM_ORDER</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 601,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 8)\n",
      "Shape of inputs array test (71, 8)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "1.062511920928955"
      ]
     },
     "execution_count": 601,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 594x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "thunderstorm_predictors = data.drop(['DAMAGE_PROPERTY_NUM',\n",
    "                                     'BEGIN_LOCATION',\n",
    "                                     'END_LOCATION',\n",
    "                                     'CZ_NAME_STR',\n",
    "                                     'BEGIN_DATE',\n",
    "                                     'STORM_ORDER'], axis=1)\n",
    "add_mean_income(thunderstorm_predictors)\n",
    "(predictors_Tensor_train, predictors_Tensor_test, outcome_Tensor_train,\\\n",
    "outcome_Tensor_test, outcome_mean, outcome_std) = preprocess(thunderstorm_outcomes, thunderstorm_predictors)\n",
    "\n",
    "without_storm_order_model = leave_out_Net()\n",
    "\n",
    "# Train one layer\n",
    "train_neural_net(without_storm_order_model, outcome_Tensor_train, predictors_Tensor_train,\n",
    "            outcome_Tensor_test, predictors_Tensor_test)\n",
    "\n",
    "one_layer_mse = compute_MSE(predictors_Tensor_test, outcome_Tensor_test, without_storm_order_model).item()\n",
    "print(\"Difference with original mse\", one_layer_mse - one_hidden_mse)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>Leave out MAGNITUDE</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 602,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 8)\n",
      "Shape of inputs array test (71, 8)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "1.1245125532150269"
      ]
     },
     "execution_count": 602,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 594x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "thunderstorm_predictors = data.drop(['DAMAGE_PROPERTY_NUM',\n",
    "                                     'BEGIN_LOCATION',\n",
    "                                     'END_LOCATION',\n",
    "                                     'CZ_NAME_STR',\n",
    "                                     'BEGIN_DATE',\n",
    "                                     'MAGNITUDE'], axis=1)\n",
    "add_mean_income(thunderstorm_predictors)\n",
    "(predictors_Tensor_train, predictors_Tensor_test, outcome_Tensor_train,\\\n",
    "outcome_Tensor_test, outcome_mean, outcome_std) = preprocess(thunderstorm_outcomes, thunderstorm_predictors)\n",
    "\n",
    "without_magnitude_model = leave_out_Net()\n",
    "\n",
    "# Train one layer\n",
    "train_neural_net(without_magnitude_model, outcome_Tensor_train, predictors_Tensor_train,\n",
    "            outcome_Tensor_test, predictors_Tensor_test)\n",
    "\n",
    "one_layer_mse = compute_MSE(predictors_Tensor_test, outcome_Tensor_test, without_magnitude_model).item()\n",
    "print(\"Difference with original mse\", one_layer_mse - one_hidden_mse)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>Leave out BEGIN_LAT</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 603,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 8)\n",
      "Shape of inputs array test (71, 8)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "1.0464245080947876"
      ]
     },
     "execution_count": 603,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 594x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "thunderstorm_predictors = data.drop(['DAMAGE_PROPERTY_NUM',\n",
    "                                     'BEGIN_LOCATION',\n",
    "                                     'END_LOCATION',\n",
    "                                     'CZ_NAME_STR',\n",
    "                                     'BEGIN_DATE',\n",
    "                                     'BEGIN_LAT'], axis=1)\n",
    "add_mean_income(thunderstorm_predictors)\n",
    "(predictors_Tensor_train, predictors_Tensor_test, outcome_Tensor_train,\\\n",
    "outcome_Tensor_test, outcome_mean, outcome_std) = preprocess(thunderstorm_outcomes, thunderstorm_predictors)\n",
    "\n",
    "without_begin_lat_model = leave_out_Net()\n",
    "\n",
    "# Train one layer\n",
    "train_neural_net(without_begin_lat_model, outcome_Tensor_train, predictors_Tensor_train,\n",
    "            outcome_Tensor_test, predictors_Tensor_test)\n",
    "\n",
    "one_layer_mse = compute_MSE(predictors_Tensor_test, outcome_Tensor_test, without_begin_lat_model).item()\n",
    "print(\"Difference with original mse\", one_layer_mse - one_hidden_mse)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>Leave out BEGIN_LON</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 604,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 8)\n",
      "Shape of inputs array test (71, 8)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "1.1291857957839966"
      ]
     },
     "execution_count": 604,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 594x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "thunderstorm_predictors = data.drop(['DAMAGE_PROPERTY_NUM',\n",
    "                                     'BEGIN_LOCATION',\n",
    "                                     'END_LOCATION',\n",
    "                                     'CZ_NAME_STR',\n",
    "                                     'BEGIN_DATE',\n",
    "                                     'BEGIN_LON'], axis=1)\n",
    "add_mean_income(thunderstorm_predictors)\n",
    "(predictors_Tensor_train, predictors_Tensor_test, outcome_Tensor_train,\\\n",
    "outcome_Tensor_test, outcome_mean, outcome_std) = preprocess(thunderstorm_outcomes, thunderstorm_predictors)\n",
    "\n",
    "without_begin_lon_model = leave_out_Net()\n",
    "\n",
    "# Train one layer\n",
    "train_neural_net(without_begin_lon_model, outcome_Tensor_train, predictors_Tensor_train,\n",
    "            outcome_Tensor_test, predictors_Tensor_test)\n",
    "\n",
    "one_layer_mse = compute_MSE(predictors_Tensor_test, outcome_Tensor_test, without_begin_lon_model).item()\n",
    "print(\"Difference with original mse\", one_layer_mse - one_hidden_mse)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>Leave out END_LAT</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 605,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 8)\n",
      "Shape of inputs array test (71, 8)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "1.0543808937072754"
      ]
     },
     "execution_count": 605,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgcAAAGGCAYAAADvpSLbAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzs3XeYXGXZ+PHvzGxv6YUeDcnj+kOIomJXsNEs2LCgApYXEBu+Kgq8ilIF6QgISEso0ksKEEhPSNn0sHnSs9lkN1uyfXf6/P44ZzYzk5mdM31m9/5cV67snDnlnnbOfZ5qCwQCCCGEEEIE2XMdgBBCCCHyiyQHQgghhAgjyYEQQgghwkhyIIQQQogwkhwIIYQQIowkB0IUOKWULdcxpKLQ4xdiOCrKdQBC5IpSaiHQq7U+N8pznwMWAB/RWq/JwLEvBB4BTtdaL0w0PvP5Y4GHgAuANqXUFGA38G2t9XPpjtk85seA3wOfAmqARuAV4J9a6wMJ7ms0cC9wG1BnLgsAv9da35pkfJ8jgc8t3nucbkqp6cCvgS8DxwADwBrgLq31a0nsrxa4FfgY4AGWYrx/u9MWtBiRpORAiNwIRPyfjC9gXGSyQil1GbAMIyn4FXAWcCfwdWCdUuqjCe5yBvB9ILTk4OPArBTCXGvuo97i+pcBv0vheJYppc7FiO8TwD+AM4GLgHbgVaXUbxLc32RgMTAJuBj4JfB+YLlSanwaQxcjkJQcCJEbTeb/+3MahUVmicFdwN1a69CL2EKl1CxgIfBfpdQHtNY9yR5Ha/1OKnFqrbsBy/vQWr+byvGsUkodDTwBrATO0lq7Q55+WSnVCNyklHpKa33Q4m4vBSqBc4LbKKVWADsxko5b0vYCxIgjyYEQFimlTgZuBk4zF80G/ldrfdC867sRGKO1dprrv4Rxdz0qZNlrQD/wG6AP40SOUqoKuB34JuAnzok9pFoCoFUpdS3wqPl4ilJqDvA5oBO4V2t9fci2lcBNwHcwSgFWAr/VWq8b4pB/BDrM/8NorTuUUr/GKM7/AXC/Gd89wDcwkorjMIrPf621Xh9S/A+wWin1mNb6wtBqBaXUX4Fzze3/D6MYfhnwQ+CrwFXAKOBV4Oda6/7QagWgDaOaJZrg8RZiViuEbPsZ8/05FTgA3KC1fijk/TsF47M6DThoxvZXYKbW+q8xjneZGeulEYlB0A2AAsab+8SsJvoH8FmgAngb4/u23dymGbg9NJnQWjcqpQ4A74kRhxCWSLWCGOlsSqmiyH+AI3QlpdQMjDvSEuDHGPXGnwEWmRfbeUAZRpFxsJHdp831P2ouK8Y40c/VWh/QWldprYPVCk8D5wH/C/wE+F5wXzHMBq4z/z4To+1B0PXAKowL62vAdUqpr4TE9QrwXeBq4NuAE6MEYGq0Ayml7MDngbe11q4Y8SzCuBifE7KsFKOK4F/m8cqBBUqpiRjF678w17sI+HuM/SqMhOT3wE8x6tYXYRSjX4aRkH0f4/OI1IRRxRD679+AD+MuPpangOeBs4F1wINKqfcDKKUmYSQQ5eZrupnDyc9QvgJs0Fpvi/ak1rpDa/1VrfUW8zjHYnyG08zXeRHGBX+pWQqB1vo+rfVVofsxv6cnmHELkTQpORAj3dkYDbniuQZoJaRIWClVB2wCLtZa362U2g2cjnGHdwpGke+7GEnCYuCT5rK5oTs2SyTOAb6rtX7GXLaK2He9aK1blVI7zYd1Wutgg0SAR4J3sEqpxcC3zLheBb4EnAF8UWs931xnHrAF40784iiHGwdUA3uGiCeglNqLcWEKKgKu0Vrfbx7nHXMfl2qtr1VKBYv0N2utdxJdFcb7u9Lcx7kYF+UpWuu9wGtKqXM4XJoTGpOLkCoGpdTHgQuB/9NavxXrtWA0DrzN3GYtRtJ2FsZn+SuMm6qztNad5jptQLwGoFOANyIXmoloKL/W2g/8FiMB+aLWus1cdyGwC6ONxBHtJMzkbjZGCc2jceIRYkhSciBGuqUYRdCR/y6JWO8zwMuhRcJmffVGjNIAgNcxLrxgFOmvxqiL/5S57EvAOq11c8S+P2n+P5g0aK2bgBVJvqblIfvxYvQoGG0uOh2jWmNRSCkJGBeuz8fYX7DBoDfOcaM9/3RILK0Yr+nTcfYTKoBxsQs6CLSaiUFQO4dfX1Tm3fbzGK/zxjjHHEwozASgFyOpA+NzXRhMDEwvEf+9cRDR+NRsx+GJ+HeX+fRngAXBxMCMpQ14i8Pft9B91QDzMap+ztZaW0l4hYhJSg7ESNcVrcub2QYg1BjMuuAIBzHq7cG4uP/ErGb4HLAEo2ThArNo/kvAnCj7GAN4zMZ0oZox7tgT1R/x2M/hG4FxGPXX0eq9Y11QWs19Hh/nuFMwuySanBEX0eC+VJz9hOrXWvsilyWwPUqpEozEYAD4YUhVTsxjRjwOff/GY5QgDNJa+8zSg6FElqqA8d34SMjjV0L+HgOsj7Kfg8D/i7L8MjO2T4YmFEIkS0oOhLDmEEaXsUiTMe5cwahOCGDcGX8KoyphEUby8Hngg0RUKZjagWKz33+ocamHfYQuoIXopSVR2ziYF9PZwJlKqbJo65h3wZPM9YLKlFIVEatONI+fTfdidJv8VpRkJVEHgAmhC8zEL95n9RrwYaXUMcEFWus+rfWa4D/CEzYr37dQHwbeTHSsCSFikeRACGuWAl8z70KBwQFoPoDRgh6tda+53qUYxdzLzJP1TuBajJ4DK6Pse6H5/zdC9j0Go/HdUCLvqK2+jgkYLfRDL0w/wBhMKZabMF7THZFPKKWqMXom7OPIMQrODVlvIkajwGAvhWTiT4hS6lKMhoy/iNMbw6rFwOfMYvygs4DiONvdDXQDDyulyqPEeQyHS6DA+JxODx2vwPz785jftwgvIe0MRBpJtYIQ1lyPUZc/Vyl1O0a3tOswGtg9FrLePIzuZ3Uh/f0XY7Q2fypKETlaa62UmgncYd6ZNwB/Iv4FJ3gX/A2l1BGN3WJ4FaMtxByz+2MDRvfJX3BkO4vQGNeaF9r7zIZvD2L0BqjFaBw3Fviq1rorYtN7zeShFaPL3yHg/oj4z1FK9Wqtt1p8DZaYDRDvxLhrX6eUOo3D7SdcSSYLd2EMNjRbKXUzRqJ1g/mcP9ZGZhfD84FnzFjuw6hWqMAYzOonGCUH88xNbsdoPPmmUurvZtxXm+sckaBhNCiN1ZNEiIRJyYEQFmit6zAaGxYDz2JcdJZg1PGGDvoTrDZYHLJskfl/tPYGQT/B6I54Lcbd90qMC/lQ3sJoBHk3RhfIuMzk5MvAmxhJzByMxm8Xaa0fiLPtQxh3/u0YQx6/gdHF8DVghtY6WgPKKzB6QTyJMeDTp0ISiC0YXQr/RGYG7PkyxucVHJnwHYwGkSuAF5PZoda6HfgixrnzOYxeLL81n+6Ns+0bwMkYbQsuwfh8n8QoIboeODE4hLLWeh9G9dQB4HHgYYx2Cx/XWjdG2f2LGF1GhUgLWyCQyuitQghxpJBBmiYMpwZyZmlERWhXSHO+BA18TWv9SsyNhSggUq0ghBDWTQX+o5T6E0b1zCSMkpFtRBnHQIhCJdUKQghhkdZ6JvAHjEaOr2P0hNiMMbumM5exCZFOWa1WUEpdgFFHGcDoS/yryD7mZqOhezBGRjsAXGAOCIOZrf8Yo8RjJnCthT7LQgghhEhA1koOlFIKo9HRmVrrGRgtvV+IWKcEo5HPb7TWtebfD5vPnY0xUcypwEkYI719O1vxCyGEECNFNqsVXMBPg6UAGEOiTg7tN44xEEu31jrYj/dh4PNKqXEY45s/aQ4c4sRo7DRUv2whhBBCJCFrDRK11nswJ24xZ4a7DXglYvrS4zAGUglu41ZKtWJM1XocRtetoEbg2Mjj1NXVSTWDEEIIYdGpp55qi1yW9d4K5rjzj2Jc7M+MeNpOxOQkGIN/+KI8F1x+hIqKyBFbU+N0OikrizpqbEEabq8Hht9rkteT/4bba5LXk/8y8Zr6+6NPVZLV5EApdTzGwB/1GK17ByJWaQCODlm/GGPM8v2Rz5l/RxsMhNra2jRGDfX19WnfZy4Nt9cDw+81yevJf8PtNcnryX+ZeE11dXVRl2ezQWI1xhjyL2itvxslMQBjVLhxSqngBDAXAyvMyVJeBn6glKpUSpViDC36UuYjF0IIIUaWbJYcXI4xZel5SqnzQpafgzGT29la6wNKqW8A95jVD+3AjwC01q8qpT4ArAJKMJKFx7MYvxBCCDEiZLNB4o3AjTGenhGy3irgozH2cQOHJzkRQgghRAbICIlCCCGECCPJgRBCCCHCSHIghBBCiDCSHAghhBAijCQHQgghhAgjyYEQQgghwkhyIIQQQogwkhwIIYQQIowkB3lqxc52rnpxU67DEEIIMQJJcpCnvvfgO8xa2ZDrMIQQQoxAkhwIIYQQIowkB0IIIYQII8lBjq1t6GDRttZchyHy0Ob9XRzqc+c6DCHECCTJQY5941/L+fF/VuU6DJGHzr17KV+5e2muwxBCjECSHAiRx/Z3DuQ6BCHECCTJgRBCCCHCSHIghBBCiDCSHAghhBAijCQHQgghhAgjyYEQQgghwkhyIIQQQogwkhwIIYQQIowkB0IIIYQII8mBEEIIIcJIciCEEEKIMJIcCCGEECKMJAcW7TvUz78X78x1GEIIIUTGFeU6gELxo/+sYndbH1//4DFMrC7LdThCCCFExkjJgUW9Lq/xRyC3cSSj3+3ls7csYPWeQ7kORQghRAGQ5CAL2ntdeH3+nB1/y4Fu9rb3c/PcrTmLQQghROGQ5CDDnB4fp143n6te3JzrUIQQQghLJDnIMJfXKDGYs7kpx5GIbPjFk2t5dNnuXIchhBApkeRAiDSavbGJv776bq7DEEKIlEhyIIQQQogwkhwIIUSO3TF/G1+7Z2muwxBiUNbHOVBK2YBHgU1a61sjnvsRcEXIolHAscCxWuuDSqk2oDHk+Vu01rMyHLIQQmTUHfO35zoEIcJkNTlQStUC9wKnAZsin9daPw48bq5bDCwGbjITAwUc0lrPyGLIQuSltQ0dfONfy3ntl5/ipGNG5TocIcQwk+2Sg18ADwENFtb9I9CitX7AfPwJwKeUWoJRovAccL3W2peRSCP4/AU4+pEYtt589yAAi7a1Dpvk4N0D3bx3QiVlxY5chyLEiJfVNgda68u11k/GW08pNR74HfDbkMVFwHzgTOAzwJeBX2YizmgO9bmzdSghRpyWHidn37WEP794RIGiECIH8nVuhZ8DL2utdwUXaK0fDF1BKXUb8CvgjsiN6+vr0xqM0+kc/Hv79u20V1h/23rdRsGG3+cfMq5Yz6Xjtew5aMTfP9BPfX09Tqcz7e9RruXba0o1ltDvXLR9tbcZQ2G3trZQX+9J6VjZEO/zaewyku+VO1ry6nMcSia+c7l87fn2G0rVcHs9kN3XlK/JwfkYF/5BSqkfAhu01hvNRTYg6lmxtrY2rcGEfhjTpk1jYo31iZe6BjzAXuwOe4y4jPznyOdiLU9cT9kh4AAV5RXU1tZSX1+f9vco1/LnNaXncwv9zkXb17i9W4FOJkyYSG3tiSkdKxvifT4lrb1AI6UlJXnyOcaX3u9c+n7vycqf31B6DLfXA5l5TXV1dVGX511yoJQaA5wILI946iTgm0qpbwIlwOWA9FSIYcO+TsqKHajJ1YPLbLYcBiSEEKJg5HycA6XUh5VS60MWnQg0aa0jSwWuBQ5h9HLYiJE8PJSdKBPj9wcIBHLbgPFr9y7jy3cszmkMQgghClNOSg601heG/L0GmBHyeDVGghC5TT9wcTbiS4Xb62f61XO57HNT+cOZ7zv8RAF2dujoc1Ne4pDW40IIMcLkvORguHF6jQaIT6zYCxR2Uf4H//4m33lgRUb2vbW5mylXzubtrQczsn8hhBDJk+RADGljY1dG9lu3twOAN99tycj+hRBCJE+SgwRlo3ag1+XNwlGEEEKI6CQ5yJIel5dXNxywtO45dy3JcDSi0BVwbZUQogBIcpBFL6+3lhzsbe/PcCRCFA631z84XLQQIjskORBC5LVbXt/Kzx5fw4qd7bkORYgRQ5IDIURe23doAIDOfpnfRIhskeQgQflQ19vY0c/S7W2W18/1gExCFJKD3U6au5zxVxRiGMu74ZNFfGfcugi3z8+em85JaDtbXqQ2QuS30254CyDh35cQw4mUHGRIJu/V3T5/BvcuhBBipJPkIM3k3lwIIUShk+SggHh9/mHTfmCYvAwhhBiWpM1BhqWzJOHEq+Zy/oePi7ue21s41Q6FPPeEEEIMV1JykEWeJNoKRJYUPLNmX9xtNu3vTPg4ojANl5KkoOH1aoQoXJIcZNGiba25DkEME8ESl6YuJ2sbOnIbTBpIAZIQ+UWSAyEK2KyVDXzjX8tzHYYQYpiR5CBFczc1sWBreqYdvnfBDs69WyZdEoXvqVUNHOqTEQ2FKFSSHKTo0llruejR1WnZ1y2vazbv707LvsTI89lbFvDJm97OdRjsaOnlTy9s4pdPrc11KEIk5bY3NOuGQXVdKiQ5sODVrV25DiElw62RV1PXAHfM3zbsGuOlam97P/s7B3IdxmBvmfbe9JYcyKctMqXX5eWMfy5kY6PRmPuut3dw3givrpPkwIJ/rRwms8ENk1Zfl81ayx3zt6MP9uQ6FJEF0t1VZFrd3g52tfZxy+s6p3F0DXhyevxQkhwkIRAI8NrGA7i8viHXEZkx4Dbed3/hDOcghBBDWr3nEKdc+wavb2nOdSiAJAdJWbajncufXMfNc4/MMm0FepvT2uOytN4Laxv5/bMbAKP4+KElu/DKXA/D1gLdwrMWxtYQIlKP08P6fTLmilUbzPdq5a5DOY7EIMlBggIcLvpp7o5fv1sIycIbW5r5yPXzLU0DfcV/N/BsXSMADy7ZxXWz63lyVUPCx5RylcJw0SOr+f1zG3MdhhjC1uZuuvrzpzg66JKZdXz93mWDJX2isEhykGbLd8S/wGZaojUadWar3E37E2t42e00Tkh9ruR//PmfOgmR3868YwnfvD/xxnObGruYcuVsGtr7MxAVbNhnnE+8Cdb/Pb5iD81dzgxEJBIhyUGa/fyJuqS2S2Zo5e0He9huoVGeNH8wGhwFq07ael2cfutCdrf15Tiq5NkkrRIhdrT0JrzNf83qooXb0jNOSzo0dzn5v5e3pK17uEieJAdx9Lm8KW1v9RT+0rr9UZcPdWH/4u2L+eLti8OWLd/RxtOro9cR5/pycuOceh5YtDMnx/7mfcv5yt1LAZizqYndbX38Z+nunMQihIguWMrQnUet9kcqSQ7iuO3NbUlt1+f2JdRjwesfel2r9Xbff2glz5ltAiLt7xxgypWzWbg78buMaOa/ezCh9R9YvIsb525N+DjzNjfzs8fXJLxdpObu4VNUGZBWG5bsbe+jvmnkDizm9weYs6kJf5zzixCRJDmIozOFhj4daWgk9PvnNjJr5V7O+9eylPfV2GE0oFywKyI5SPK8sbM1O8Xyl8ys480oiUi+XSClZXb++ewtCznrzpE7JPmsVQ1cNmvtYBXCcLNiZzs/e3yNJD8ZUJTrAIazdBTjP7+2kefXRi8JSLcC6FgB5G8PkK/fm3oCJ/JHc5eTSTWlOfu+7WrtxeX1U3tUTULbBQIBbn9zG1+dcTQtZmlZi8WuyoXm54+vocflpdftpaasOGPHeXn9fipKivji+ycdsdzl8fOdjxyXsWPnipQcDAMbGzvZ2x77Lt7qHfZNc7fi9GS+21Hd3uz14+13eznnrvTfOe5p6+OmuVuHrDqat7mZTY2FPfR2tuTjfd/HbnyLh5bkrl3KGf9clFSpR0e/h7ve3sH3H1yZgahGpl8/vT5q1eavn17PH54fnl19JTlIUOg9RL70AvjqPcv47C0L07KvLQcyXz/7zftWZPwYQev3dcZ9TTtbeznzjsV09lufC+Cnj6/h/kU72TNEN7BLZtbxlXuWWt6nyH2j2UjLd+a+a3KigglrvHZMQgxFkoMk5GmpdkzBiXDyyasbDuQ6hEH3vL2Drc09vJ3A1Nv5Pipkn8vLda+9m5WSoGzJl2RciJFAkoMR4MJH0t9nuN+dWhfPVbtzP0RovjVoTKd7F+zgoaW7mfnO3lyHkpBon0ihJeMinJVf2f7OgYIspRnOJDnIgZZuJ6/k0Z1zMpfIg92F28Ap8lqT6UmydrUm1nW0ucvJr15LrRFqsEjZVyBFy5GfSX1TNw8u3pWTWER6JJLTnX7rwhHdRmLLgS7mbGrKdRhhJDnIgQseXsmvnlpHb4oDLA0H+XRXmEgsiVxyz7wjsUZlj63Yw/b2ods/DPcREs+6cwnXz6lP+36fr2vMaoNYgIeX7mbKlbOHnMV1pMvHqs9sOueupaxtyK+u0FnvyqiUsgGPApu01rdGef6fwLeB4C9Ya63PN5/7E/BjjLhnAtdqrQvj1ihEU6fRvcifJ5Wo6brMdPa7qSgpoqQoOzlnPrx9Vt47d563TxgOWntc2G0wrqp0yPV+Z84ouuemc1I+5uI9vUw+3s2YypIh1/v7a+8C0Ov0UlrlSPm4QmRDVksOlFK1wFvAt4ZY7RPAd7XWM8x/wcTgbOA7wKnAScDpGElERuXTnW2+m/G3N7lkZmJzS8x8pyHhkRbT/ZHkQY4hkuT0+PD4/Hzk+vmcet38rB23qWuAGxe1cOms5OZSybZ8SKRHug37OgtqMKpsVyv8AngIeDbak0qpUuCDwB+UUpuUUs8rpY43nz4PeFJr3ae1dgKPABdkI+hUWE0uhksOkkiL/6CfpmFo5JgSOCkmU1QfnJky32S6HUU2tfUYVSyvbTyync77rpnH9x98J2PHjvUuBovBD3Tm95DcmTqvtPe6eHFddgZnGy6+du8y/lBA059nNTnQWl+utX5yiFWOBt4GrgZOBt4BXjarIo4DQtOuRuDYTMVqRf8wmKf8yZUNw3b0tEwKXnu/es8y5m3Or4ZEoXJR8tXel97v0/YWY+bRuZuboz6/ek9HWo83nKSaIu5u64s6+dylM9fy22c2sL9zIMUjhBtGOW3S8qUXVV4Nn6y13g2cHXyslLoVuAaYgpHIhL5rNiDq1bm+Pn0Nmbo6wxuJbN++nf2txt3Com2tQx5z27ZtlDjCz8719fX4/EbY2/ThSZ2amlK/wESLIdoyn89HfX09Lb1e/vxiQ9hze/fsoaI//CQc3Efd/sMD/uzceXh2xZaWFurr3Uesn2rssdZxuoz3f/fu3Ti6jTpmp9MZdR97m8JPXvX19TQ3G6MWdnR0Ul9fT1eX8Xj/gf3Ul1kbBMrtPvx6Z6/ZwQmO8O9JZCyJvCftbeEN5qJt2zbEOvX19bS3G88fPBj+2SQrkfh/+JjRy8Dlch3+zGJ8PkGNXUaMbrf7iNfi8/nCHicTZ7KfR29vb9R1D3R7osY7lG3btzO6bOg2B1Z/w0Nt2+k03i+v10tbm9E9sK2tlfr68It8R4eRVDU3N1NfPxD1MzrrsV2cNLGMW846Omx5Q5vxO9mqt9NdE33Y4tDzXGWJtfvQnTt3AODxeCy9bu8Qx4j3nQu1zzy39fb2HfH9C2X1+zWUeOseOnQo5jqJvKZU5VVyoJQ6GThFa/1EyGIb4AEaMEoWgo7GKD04Qm1tbdpiGrXZDRzuijZt2jS6SjqA8OLz2tpa7lsYPh3x9OnTKS22A3vC1nPYGwAv09X0weeOOuooILV+vsbr3hV3mcPhoLa2lsr2foy39bATpkyh9oQx5qNdIfuA/3l1weB6U6dOJViQM3HiRGprpx6x/pFid02Ltc0zqxuOWKfsjTbAzXve8x5qjxkFGD+4aPvoKGkDmsL2saZzD9DO6NGjqa2tpWa9E+jj2GOO4egTJnLxY6u54/wZHDe2Ima8JSXNgHGyHTtmzBHvc/THcPadSzhubDkP/PDDMfc9bu9W4HCyEe11jW/QUdY5/P6P2/Uu0MWkScHPJlnGPk+YOo2KEqunC2Ob0tLSwdhjfT5Bpa29QCMlJSWcMHUaoa/F4dgH+Acfb23u5ldPreO5Sz9B9O/7kbEcXh7vO3p4HYCqqqqo61a09wH7KCkpsXC+MfY3fdq0IRpMRovLSqxHrtfe6wL2UlRUxPjx44FOJkyYQG3ttLCtxujNQDeTJ0+mtnZKjM9oF5tbnGHL39jSTHOvccwTT5zKCeMqo0YVep6LP++Bsb+pU08E9lFcXGzpPB48hlLTqY44RrzvXKhWRyvQTKsTjntv+PcvNL7436+hDLXu4e/c2LFjw9ZZvecQ/+/oGipKihJ6TVbV1UVvN5NvXRn9wF1KqfeYjy8FNmqtG4GXgR8opSrNtgkXAi9lO8ChCnxunhd/OuJ86ndeCI0t//j8pvTvNOKFh34ir206QN3eDv4Vkeily7tN3by+JbEGmPkgI59DDCf95fUhn7/tjW1sO9jL8h3taT3uB/7yOv9enJnPPR8Fq2sS9Z9luZtvItMaOwb4QQbbsCSqpdvJt+9fwe/+uyHrx855cqCU+rBSaj2A1noz8EvgVaVUPUYjxO+Zz70KvACsAjYDdcDjuYh5Q2Py/VHvfnt71OW5SBnsdmvZwQfinKyH4vb6aepKb71kpmSyvrMrDdN359K25uQuJMmIzJ+z9dvocXm5YU78BH+4eGdXdsZ7KLR2BBtyPFlaaMPoPrNdW31T5ue8iZSTagWt9YUhf68BZoQ8nokxhkG07W4Absh0fPE8sMjayG3r9nVwklnsHRRrlr5nc9DFZajU4LY39ODfPUkO1hQIBFDXzCUQgG3XnZXUPrJhf4dR33jH/G387DPvtbRNoo2GTvnbG2npW29VoZ2QR4J8/Ui27O+mayADyWuME8xC3cLe9n5+/Ikp6T9mFtQ3dXP0qHJGVWRuiuh8kFdtDoabix9dw0emjIm/IuRkdCz7EPUKd729I+X9L9neNniR8vrzdyCg4FDQQ82wmG+sd5EtgLqjPDYSkqxn1uyjvrmbmz8/LivHC871UojJgcfn56w7l/C+ydXM+81nch1ORuW8WmG427Q/t0VUQ0mlzcEbW6J3KwuVkbsREow76ZP7kRs2tPfzmX/6qQYHAAAgAElEQVQs4GB3fvdtFyJSvCRnY46L0gvFrWaJ6tYsVrPliiQHcWTyzsHry91tycWPrub62cl3iblx7vCsmx3qTvsztyyg4VA/r6yPP2nWGf9cmEIMIlQ+vh+FUqKQ6fdu1e5DbD+Y+wtlsh+Hzx/gG/9aFtYtfSj7Dh0uXYw3R8fm/V08GqXx5v2LduLPo4bpsUi1QhzPr83cKGDfum95xvYdTzIjGSbq3gWpV00Uql2tfQlvs7e9jx0tic3gmE6b93dx3NgKRpXnV11qPp5Gg/OiZKPHT31TN7VH1aS0j9vnb+NHHz8h7jwQifq9OeJfNtvThEr17e/sd7O2oTOpat3r4txcnXv3UgAu/OR7wpbfNHcrx4wu5yunHB1ts7whJQcJauxIX730rrbELyCFJN1Fb1OunJ3W99/qRScQCGRlmOTP3bqQnzyWwaGk4zj37qX84KHsduNyenzU7c3fEQ5jXfzvfMvoddSZhV4o6ert81aabgiSbccSCATyqit3qtal0E7MVQCzUEpykKBn1yRWkiANwtIr4e5XUd7+RD+Rx1fs5eS/vpHgVolLZ1F1srvavD9+l6mufg89aUqW/vDcRr5533Kau6y342gy1z3Ul/rIj8l6dYNRtZSpdjXD0UWPrmbqn+cMPj7jnwt5ad3+HEZUOHLS1T0HxxxRBjyFP/9CUG+SXRrzXbxJit5McNbI4e6Uv73BB//2Zlr2teWA0RAu+N2yMmFUsJFvohcWTx5One30+Oh3Z+53lU/36Qt1eL3+rtY+rvjv+hxFkz1+fyDhErkpV87m8ifXDt7I7G3vz3oiKslBmlxnztle6PyBALM3HjnPw1Uvbsrqndrahg7LjYQSZeWEmchdvC0Ph5oMRuT2+XFmIEH1prl4OBvv4Ytrc3uXGu079bEb3+L9/5f8IGMFIQMZyoHOAc66c0nSY7Ck27Sr5rAxxuB4XQMeliUxmudrEefhy59cm1RsyZLkIE0eWjo8hhR9ZNlufhHlSzhrZUOUtQ2ZOK9/41/L+fF/Vg25jqUL+BDrRAvb6msplNbqt7yued8181LeTx7mPwn7w/Oxp8uNVWKR6c/ZSpuFfPuuWR0ALJNfmcdW7AkbNTDXX0+PL8BX71mW0WPszfI4LNJbIUHpOEnm2W89TCJ1v4lI591Rri9UuZ5S1e8PcHcaBqkqFLEujun8HPLtAmxFS7eTfy+2Nlprocn1b1xIciBE3onVkjkQCKAP9tDjzH5RqpWLZ0N7P3vak+uBY6WtQaJahslgVbEulFe+sCkrXZIjDdXI2unxcedb2/nVGdNiriMKgyQHIi3ueiv6hFLp9nxd+sediHdZyvZdTOhAK6EeXrqb62bX879fmh53H1YvtT5/ALstPXX+n7llQfyVQnQ7PezvzNykXB+94S3L6xZgwYGlBpbZvgGf+c5e7lu4k2KH1FgP5R/ztnLl8xvZccPZuQ4lJvkEE5TqqIb5fhJK9iJx25vb0hxJdO/sjt2w59k1+5hy5WxumGNt5McAxsXRHeVOPd6NbC6KoTebrfQbO9J3QZ365zn8/Ino87mHSuZrEe+7dO5dS3F68qMHQaIlF0Ot3TXg4f5FOzNSGpLvgqVe+dgzJJ+09LhiNurNdbVlkCQHCXo2DXeu0S5G+aIQTmixQgyO1havHjb0kvXzx9fQHtILIx+qOrNdUhGtq6bb6+fqlzZl9LgNMUpICt1fXt7MTXO3snh7W1r21+fy5XzwoF6XNyNd6fwB+ORNbye0TUu30/LMuCJ5khxkmdPjw53jrHqoBKDhUOaKedMhNPRdbfGHGo7X1Sldo8YlKxAI5OVd1tzNTcx8J3YPFSvyKdGMN5Z9OiMNtgnxpOkm4JdPreMPz8XuaZEpHSFJ86l/f5NTrrU+ENh9C3daXjfRqqULHl6Z0PpDycY3NF3HaDjUn9UkUZKDLFu+M/H+rtnU1utKeR+LMzQ+QaRfPx1/AJX/iSgy39jYydUvbba0f58/gMubvjECdrYemcw8uGQX066aG3Yijt06P3usXtcTHUNhf+cAT7yzN4E4En/Vy3a0MeXK2WHLNh8YetbBeIfZ1dpLX0iimUzec8V/18dsTxJPuuZ4CQQC3DxvK3ssDN3+oesOD3RldbjfbJR6bTsY/aZg3uZmbs9S9WauzNuevUmuJDkQafej/6xigc7uHXmnM/wiFeuinsgd2GWz6lBXpz5GQFC02euerzMG5jnYkz8t661ekC+dWcf7rpnH+n3Wx5j/4cMruealzWHJEMROfOZubk74ghOtmiTZQozgZmf8cxEXPbo6uZ2Ylmxv488vpq+qJpn2QXvb+7lv4U5+8lj81xLtPZty5WxW7Mr8DU4yn9clM+sG57wItbGxkxV5flNm1UAW2+hIciAyorUnsRIIK3cyEPvO5PoF4ReEfTGqR/wWzzoBAry+JfqwyXlUWh5TKjFanTBr7uZmANY3xJ44yenx8U7IxSQ46I/Vz+Evr2yxtF4q1jV0WGoEtmp3gvN65Mip180f/DvyVQUfxyqefnpVA9OumoM3jVVdAXI74dJX71nG9x7M7oRiw4EkByJlwbHuU/FIlHnPY4lMELYc6GJzi7U773gnKRl8JfUeOaH+8vIWvvvvd2jsSm7o7USTzFj63dFLkuZsauK8fy3nuRQaGlstBSkEf3/tXTy+AM4k2ks0dvTzj3l68HGwZCMQiF2Sl23pmjAsEYmeUvJlsj5JDkTKos3FsEhnp90BwH9X77O8bugN65NRhoTe0RK/kWOmvRFjoqdslVjUN3enrTvVVrMqpdc99MVmqNeW6OtevvPIXgIzY7Rz2G2WWO1siV5yZaX9zGMr9kRdnu1E8wcPvZPTmSrnmSVJQfmWaK/afYgP/PWNsCrPPAsxr0hyMIxENsLKpdmbjkwY8oEvzpXmwSX5P0dGMifdRBr2/eG5jbyQo0mK0jEYU7QGa/GSnXzpW36w25n0XfayHe08tSo84ZWL32F1e43qr3cSaDMx5crZdPbnLuHKJUkORqB8nEUwW6zWdYdK3/uVu/f9PX+ak9BFR1tsdzCUXa2JDKWc24uz1cGYvD4/d8cZDTSVIY1Pu+EtLn9y3ZDrtPW6uH72u3G7Zw4lP1KhwrAzoe9xeuXyVC3JgSg4e9vidweL9aPyW6xKzUYRfku3Ex2lB0MsqYbU0We9vrUjDXdLmR7PI5HPKN66kXfcsby0/gD/TLG73PaDPdzy+taYpTnReluEuurFTTy4ZDdLdyQ+yNLIvS3Ijmi9Igo1EZPkQBSUQCB8YCOvzx/1zv6Z1fuith9IVwO3dMjkzIqpFpN7EmiUmM6TXzIlO8kIBALsO9RPY0di4w6ko2Hd9x58h3sX7KTDwnTN0aSzwahIr2i9Ih5YbH1AqHwiEy+JvJDs6e5918zjgo+dcMTyfy/eFXUY5WTuZiPv8CJj7YszCmO65NOIg6mKNXtjtuasf3DJLm6YszUrx4qUbLe+QCDAku1tQ/5WbnldD/FsZg1V/ZZoC/x0FKcv3N3Lse/Jfu+ESIU61LMkByNQPl9kFmxt4enV1oftjTV5SSZFjjHfbbF7VPrqD+PvKN3doYYa+2B/xwCn37ow+pORiZX5eHuMUe6i7yL6Z7wjyoiTscyNaEm/clfmxyxId33xKxsOWBoVNB9lu8Gnbu7h5sUtbGjP/rDTM9/ZyxdqJyW9/X8S6NadSVKtMALlb2oAFz26OubgQ7FsbLQ+Ql8yIu+I7lt45Ix7TV2Zn5MieMS6vcld2AIEWLPnUNLJYaxGig8t3T3YJTCWyOtkOi6cnUkWy2fS0jRNthTNgc78GUUzGzbv7+Lb9y9PeIhugH63UZrX1H3ke/bAol2DjTkzcS68+qXNKY+mmQ8kORiB1uzP78mVErW2If3JQSInjUAAPn5j/JnlQq+HczY1xRyYJ55YY8vHM29zM9+6fwVPrbI+LkSoL9+xOKntrBiqAKiQetfc9fbQPRms+muGRoYMvs3ZLDxM9lB/eWULq/d0DE5VbtXzMQa0Cv0apaPB7VDS1f1xl8WRYzNBkgMh4sjERFKXzVqbtsl0rApOkRyrvt+KH/9nVbrCCfPEitiTMWViquBsuPjRNUk3YHx0+Z60xPBYxPsaHAwqE9NlH1E6lMK+GjuSv4H53bMbUjhyNIfTG3eaZtq0qrkrd6VF0uZAiCz5ecQMkaHSPfZ8pup4F2Voxs1UEpZ8Ejn/wt1vZa5HSjzdA54j2uRsiTM7pThSaCnL/PrsTiiXyzIzKTkQeSHfCo7jxRN6yk1HqfemRgsn7XxuLCKiao+cdyGLn2EuGuvGkkoD2W/dv4Il29OblCbyzqQy70a6ZfMTlZIDkRfy5zQ2MqzZcyilottEHTE7YI4/8Ew0Y7j2VWvtBJId3yBVs1buxZGh9huBQIB5W5qjPrdqdzvvP2pUSvu//Ml1nHzsKP7xrZNT2k8y8mlslGyS5ECIKOJduwaSaEGdrCv+u57ffmF6yvsJvXv71v0rUt6fFRuslIiQ/WQhnccLvq+PLNuTYAzZfdHrEmi4m2hsL68/EHNK60tmrk1oX9F0DXhYsr2Nu2JU00Q2Wo0XfXDtXE5Ule+kWkGIKOI1BMpmN7oX1u7n0/9YkLXjZUK2qo3uXZC7Ov5oIm/Um0O61mW7QaoVyfYMiXZ3nasSEoDbzSGuQ1/NTXMPD3oVrHH50u2Z64Fj1WWzYrdFyiUpORAiimTGrc+01IdENlpaRxs5Mlsyfa9sdYTA5ij937MhdEjvREdovHlebkZ0zCdW58BYEme8iTNuXRg2DHsuzdkUvTom17KaHCilbMCjwCat9a1Rnr8A+D3GOaQf+JXWeo35XB1QDgTLgWZprW/JRtxCFJLn6xqjFps/8U7s7oLZluuhC7Yc6E5qu1y3lci03jy5YGZaIonBMP/IY7KcHCilSoDzgOnAPcAHgK1aa0t9O5RStcC9wGnApijPK+AW4ENa6yal1NnAC8DxSqlKYCowQWtdmJ2exZBeXLufv33tpFyHkTO3z09tpr9Qv3t2Axd+YkpC2zy50vqQ1ZmQqzv5TBtIcqCrXCukgaeG4snwzKCpijeGSi4/BkvJgVLqvcB8jDYKxwCzgMuB05VSX9JaDz0BueEXwENArLOQC/ip1rrJfLwGmGwmJR8FeoF5SqmJZix/1loPr6H+RrB8KeLLlYPduW0R/ecXj8jXRRSJtlx/cd3+DEUyPKW7ZKatN797GsQbdv2Pz+fud2m1QeJdwFzgvUAwxf8e8DJwp5UdaK0v11o/OcTze7TWs2Gw+uE24BWttRuoBhYA3wY+AhwP3Ggx9hHtrrfSM5xrNuRyNLBCMJyKtAv1tdxZQL8nIVJhtVrhU8AVWmu/UfoPWmufUupGIK1jVZpVCI8CxwFnmsd6BXglZJ0bMKocfhNtH/X19ekMqaDd9mb6iqsz7WM3vpXrEJLS0xN7xsJ06u6OXU8e+Z3v6Mj8rIOJcLnd1NfX4/MZxexNTU1xtoguF7/teMdsbWvlxufbsxRN5jidh5NzrROb+vlgS2KTpcWyY+fh3iYD/daHeN66NXq8Xm/qJZK9vYnPZeLxWKv9TvR34PV4svYbsJoc9AGTgcgrzfuBtM16o5Q6HngVqAdOD1YbKKW+AnRprYP9TmxAzHe/trY2XSEBhTkXt8ie6upqjPazmdU8ELsC0vjOH/6ujhkzFkiu0V0mlJaWoJ019LiMOuCjjjoKSLxHSOTrzIa4xyyr5oF3cttmIx3KysoG/zZuAvdY3nbSxElA6gnp1KlTAWNisHdbrVcJKDWdaPEWFxUDqbX7qKqqItHfd1u/tWMm+jsoLi5O8/UN6uqid6W0mhw8ANyvlPo9xoV5ulLqs8ANwIPpCFApVQ0sBB7TWl8b8fSxwDXmMd3AFcAz6TiuEIViZ2thzz/wm2fW5zqEjMjzNm8F5YW1+ddGI5ODVa3e05GxfafKanLwd6AHIxGoAOZgpIm3ATcle3Cl1IeBh7TWMzAaOJ4AnKeUOi9ktc9jJCfvBdaaMS8A/pbscYVIp1TGjRdiONidpomzNiU4PXM2ZLJ5TD7N2xDJUnKgtQ4AtwO3K6VGAUVa66Qq2bTWF4b8vQaYYf59I0M3Mvy9+U+IvLKhMW01a2nz5rvpqQPOlAJtjyhiyHVX2FiGSY/MnLDalfFHUZYN/q21fjyNMQlRUJryoJdF5Kx1+zull2+2yAUofVze5NoHxBqXIR1dGRfqzExTnu+sVitcH2W78Rj1/ysBSQ6EyKEfPrwq1yEMLaKo4OnV+5LazT1vS1fC4WzZjvT2+vD4pIwqWVarFY6LXKaUqgHuJ81dGYUQw9+GfclVxdz6RuF0zRUi3bKZ6iQ9K6PWuhv4C0bPASGEiGl/j4x6XkiyPZ10qroG5PuVbqlO2XwqUJyOQIQQw9ctS0Zmva3IjkueyM9pjwuZ1QaJCziyRKMG+CBGd0YhhBDDRH+BTRi1szXxUQzF0Kw2SFwa8TiA0RjxKuCNtEYkhBB55K+vbMl1CFl3z9s74q8khjWrDRKvyXQg+er76+fy6d3ruPS8P+c6FCFEDjy6fE+uQ8i6fJ/qWGRezORAKfUfqzvRWl+cnnDyz9T2Rs7attyYRk46NAshhBgBhmqQWJzAv2Hrfa27AZj5zNU5jkQIITKnzy2lBeKwmCUHWusfZjOQfPWXL1zC/Icv41N7NzDl0H72jD0m1yEJIUTa3bfq8OyAhVZIWmDhFgSrvRVswNeA/wc4zMU2oBT4oNb6rMyEl3s7xh/Pk6ecyfc3zGPhg/8DwP+c92den/6JHEcmhBDpE5xOWwiwPs7B3cDTGAnC/wFnApcCfyTbk6vnwJ/PvJyu0srBxw+8eAMTe9I7zKcQojDJXWvu9RVY18tCYDU5OB+4QGv9UWAn8FPgOOC/WO8OWdBO+c0zzPjVk/zlC0bpwap//ZhqV3qmKRVCiFyTJCf/ub3ZG7nSanJQAwRndtkInKa19mBMyHROJgLLR53lNTx26ldYdsLJAGy64/wcRySEEOkR2s6gwEZPHjE6ndkrIbGaHOwCTjH/fhf4aMhzo9IaUQH4wXdvGPz743tl3ikhROF7Z19/rkMQecRqcvBP4Eml1HcwqhIuVErdDTwGLM9UcPnsw5c/AcBTT1/FB5pkGlkhRqqOfneuQxAi7SwlB1rrh4BzgV1a6y3AN4FjgXXAhRmLLo+1VY7h6i9eCsCrj/+W1x79NZ/fsTLHUQkhsm3OpuZchyBE2g01QuIvgKe11u0AWutFwee01nOAOZkPL7/N/NA57B57DLOeuZqTDu7k4ef/DsCpl8+kvXJ0jqMTQojkSJuD/JTNj2WongZXA7crpd4EZgIvaa0HshNW4Vg2ZQbqiuf525v3c/6mNwGou+eCsHW+f/51LJ8yIxfhCSGEEAkbqlrhaIzxDPYBdwAtSqknlFJfVkpZbaswIriKS/nj2b9myh9f49wf38HOiFEUn3zmavbcfC6jB7pzFKEQQghhnS1gofxIKeUAPg98GzgP8ALPALO01quG2jbb6urqAqeeemra9jflytkpbf/jule5dv4Dg4/fnfgeNk6exgsnncGq405KNTwhhBAjxDmqhnsv+nRa91lXV8epp556xDAXlpKDUEqpIuALwDcwxjjo01pPT0uUaZBvyUHQtze+yS1z7wxbVj9hCrWte/j6D//J+qNVWo4jhBBieMpmcpDM6IYOoAxjNkY7IANyW/DsyV/k2ZO/iMPv49O713LF0lmc3LwDgJee+B3OohIaRk1mensD7058D5snTaXC7WTWB8+ivWIUNkCPP6HwZkQRQghRcKxOvFSBUUrwTeBsoBujWuFcrXVd5sIbfnx2BwunfoSFUz+CLeDny9tWcP9LN1LmdTO9vQGAco+Ts/UyqtwDnKuXDm7bWjGa/aMmMLG3g3VHK0p8Hmpb9nBsdwsuRxHzTzyNlcedxPxpp3GgZmKuXqIQQogMSLSkPxVDdWWsBr4CfAv4MuAGXsBoc/C21lo6u6QoYLMzT32SKX98LerzX9i+kulte7EH/DiLSlCtDUxrb+BQxSjO0ctorRzNhL5OXp/2MSb3tHPavs2co5fxt/kP4CwqoaOsmnVHK56acSYbjpqOy1GMq7g0y69SCCFEoRmq5KAVo8pgLvBD4DWttQwFlkXzp53G/GmnWd8gEODjDZv4zsY3mNDXgT0Q4Oxtyzl72+FBLNcerWgYPZkKj4vjOpt55uQv0VQzngPVE+grKafK3U+l24k94OfYroP0F5exbcIJNFeNo7uskoBNOqoIIcRwN1RycBnwvNa6K1vBiBTZbKw44WRWmBNDAUw5tJ9T929lbH8X4/s7+czutXy8YRMDRaVM6Wzir2/9O6FDuO1FLJz6YYp9Hnw2OwGbjfce2s/ao2ux46e1cgytFaNZ9N5TcRaVUOUeYGLvITyOImyBAIcqRjFQXIqzqIT2itH4bTaq3AMEbDZKvB48jiK6SyvD2laUeZyM7+s0/va6cQT8jB7oYVJvOy5HCS1VY9k3ehId5TX47I6k374aZy+V7gFaK8fgdcT+aTj8Pvw2W2YSpUCAUp+Hsf1djB3optrVx5j+bsYOdOOxF3FMdwvHdzZT7PMybqCL5qpxdJZX01tSgT3gp6+knJaqsfSWlDOhrwO/zU5bxWhaq8awZdJUXEUlEAhwVE8b/cVldJVVGe+1WVw5oa+Dco+LIr+PalcfvaUVuB3FOItK6Sqrwl1UDECJ14PD72PALIkq9nsp8XroKymXdjHJCAQo8vuG/N4VGuP37JAbigIV85uotf5PNgMRmbFn7DHsCRl34cbTLx78u9jnYXrrXuyBAJN725nQ10EAG51lVcZFpXI01a4+ju1qodTrZnJvOx9pfJeTmnfQV1KO32bDb7NT7PNy+q7VAIzvN3LJaxY8nHTM3SUVlPo8lPo8CW3nM2N2O4wLWHvFKDrLqin1uekvLsNvszPK2UO1y5hgZsxANzWuPpxFpZR7XJR7XWH76SmpwFVUwrj+TuyBADYC2P1+alx9+G023I5iSnxe9ow5Cr/NjsdRxKHyUXjtdsb3dzGuv4tRzl7cjiKcRSW4HcU0V4+npWoM9kCAUc5eql199BeXcVzXQSb3tFHk9+MIxG7j67PZ6SmtoKe0kq6yKo7rbGZ8fxclXg/2OOOnue1FdFTUUO52UuM23gOvzY7HUYw94Lf0fveUlNNVVs2k3naK/T782I447kBRKe0VNfSWVNBRUYPHXsSBmgmU+Dx0l1bSVVZNR3k1XWVVVLn7cTlKcAT8jBnopru0kn2jJ+N2FOPw+ziqp40PNO/Aa3fQW1LOvtGT8dodHKiZgN9mY7Szl3KPi66yKsb1dzKhrwNnUSlH9bTht9k4rvMgFR4nx3YdZNv4Eyj1eSjxenAWl9BVWoXPbqfc48Jnd9BXUkZHeQ3NVeMY7exltLOH/TUTcPj9dJdV4rPZqXIPMK6/i1Kvm4DNRrHPS7HPy0BxKd1llRT7vIzt72aUs4fusipcRSV0l1bis9sJYKPc48Jmvl9j+7s5obOJUc5eTuhoYoyzB5ejGFsggN9u51B5DX6bnVKvG2zQV1w+uG1nWTVjB7qN5zAS1k2Tp+Gz2ynzuqhwu+gsr6KpejyVbic1rl4OVo3Db7Ph8PsZ7ezhYNU4jupppdjnw11kvN99JeXYAgFsgYD5uymnq6yK/pIyXI5ivHYH1a5+Kj0DVLidVHiclHndHN3dSonPw5iBHnpLygGY3NOOz27nYNVYAjY7xT4Ple4B7IEAAZuNhtGTaasYTZW7H5/NQXdZ5WBSWuF24i4qxuUo4UDNBOP3U1TMQFEpNgL0lZRT5PdxQkcTZV4X3WVVNNZMxFVUwihnL6OcvZR5XZR7XPSUVuK32fDaHfSUVmIP+Cny+yj2edk95mgcAT+V7gE6y6txO4qpcDsH3/9yrws/Nnx2Bw6/jwqPC7ejCL/djtfmwB7w47M7cDuKcAT8VHhcjB7optjnxVlcis9mp6O8xjifeVx0VNQY3xm/jypXP0d3t1LucVHqc1Pi81DqdVPs82IPBLAH/JR53RyoHk975WicRSVsP+qiuL/RdBk+aapImMdRzJbJJwKwiWlp2act4Gdi7yG+sGMVPpudrrIqeksr8NodTOjroNLtxFVUzOiBXsYMdOMI+GgvH4WNALYAOMzqjGAJQpHfh40AB2omUup1011aSW+JccJqrRzD+P4uql19HN3dypiBbib3tFPi82APBBgz0M2k3nY8jiIq3QMU+7zGSWSU0Vhz4+Rp9JeUGXdsdgcHzaqTST2HOK6rmSr3AHa/j/qJ78HtKAJs+Ox2xvV10lI1loDNNniC8jiKKPW6Ge3sptzjorl6PLvGHkNvSQWV5oW41OflmK4WPrpvC25HMd1llQyYicmWie/lzRNPw1lcisdeZJ5QbeyvmUhHRQ0d5dU4/H6aqscfeXdp3vXbCFDhdjK+v5MqVz9tlaOxBWBcfyfHdLfywQOaiX2HcDmK2THueGwEGGsmMQPFpfSVlA8mVD67HWdRKWVeF8U+H+VeF+P6Ohnt7GGUsxdnUSmNoyYOXiQ99iI8DqNkw+0oZnxfJzWuPkY5e5nU045q24vbUczY/i4qPc6Ev1cuRxHFPl/cBCiU12bnYNU4usqNZPdD+7fSXVYJgMPvp9rVhyPgx2t3EMBGpWeAGmff4DGiJT5Bfmz4bTY8jmI8jiLKPU6K/cZ0uj0l5QwUl1HpHqDI76XU5426D2dRCTvHHkt7xSjmqk/SXD2OSb3tg0nA6IFe7Pgp8Xrw2+wU+b14HEUUmbHvGnsMHvOi5MfOSQeNJMpZVIqrqJhjulo56eBOXI5iDlWM4qSDO7EFAvjMROuj+7bQW1pBe0UNlW6n+b74CNjs2AJ+XEUllHtcjHL2UojXdh4AACAASURBVOEZGHwdA0Wl9JaW019cRn9xGR5HEbvHHsNAUSkd5dWMcvbit9tprJlIkd/H8Z3N+G32wQu8y1GMPRBg6qFGRjl76SmpoNjvZVLvocHkt71iNMU+L6U+NzOaNCVeDyU+DxVmclXmdeOz2dk3ahIDxaWMa9rOxL4O87tSTFdZFQPFpfQXlzFmwPhNBhMer/lddfh9VLszM+jvUN+dUJ1lVfSWlON2FOMybyC8dgcBmw2fzU53aSWnNG2n0jNAT2klzV1fy0i80UhyINIqYLNzsHo8sz54dlaOJ/NhMliMH8BGX2kFfaUVYU8314xny+QTeWP6x3MRXTiz2qTK1c/Y/i56SyuMi5vNRnv5KMb3dzG5p40Snxd7wEiGGkdNwl1UTKnHxaTeQ0xvazDvxKG9wrijGt/fRWvlaFqqxlLj7KWpZkJSVUxlHicT+jrpLq2ku6ySib2H8NscjOvvxOMoor+4jLbK0Xjs5qnTfO/tfh8lPg8+uwOPWXIVVGHeLQO4ioyTvy0QwG+zF1YVTCBAic9rJMo5jtvu9xGIqNqzBfwU+3yWqzJsAb+RGBeV0l9SxoTeDkr8XjrLqnH4fYzr78JVVIIt4Cdgs+OxO3AVlZg3H36jBASbkQB6PXgdDrOqrhq3o4gyr4siv5/xfR30lVTgdhQxZqAbV1EJXnsRA8Wl9Eb8VuM5+9jqhN+rZElyIITIHpsNV1EJrqKSqJOT7R81kf2jonfDdRWX0jDmKBrGHHXEc6FJYndZVdLhOYvL2Dd68uDjg9XjAWitGjPkdn67A2eMZKTfLGYPFSignGCQzTbY5iTX/FHe64DNjrvIevuGgM1OW+Xhz7WlelzY813lqV2IncVlAGEJQKr7zCar4xxUAn8GHge2AQ8D5wNrgB9orRszFqEQQgghsspqmnU3xvgGNoyk4LvAz4F24N7MhCaEEEKIoGwOLmQ1OfgaRgnBVoxREudorWcBfwLOyFRwQgghhDAs29uXtWNZbXNQBHQrpYqBLwFXmMvLAVciB1RK2YBHgU1a61ujPH8OcCNQCmwEfqK17jZnhvwnxjTSRcCtWuv7Ezm2EEIIUai6XdmbyshqycEyjAvzQ0AJ8JJSagZwDzDf6sGUUrXAWxhDMkd7fgLwCPBNrbUCdgE3mU//DzAdOAn4CPAbpdRHrR5bCCGEENZYTQ5+jtHe4BTgfK11O0b1QjNweQLH+wVGgvFsjOe/BKzWWgcbH98H/MAsbTgPeERr7dVadwBPAxckcGwhhBBCWGCpWsHsjfC1iGXXJHowrfXlAEqpL8VY5ThgX8jjRqAGqI7x3MkIIYQQIq3yrSujnegNMn1RnrOZy49QX1+fpnCEEEKI/JGt65vVBol3Ax8DnuBwV8afYVQt3EtEqUIKGoDQaQiPATq01n1KqQbg6JDnjsYoPThCbW1tmsIBo9mDEEIIkXvpvb5BXV1d1OX51pXxDeBjSqngQP+XAC+bf78MXKyUKlJKjcZIUF5K47GFEEIIgfXkILIr41xzecJdGSMppT6slFoPoLVuAS4CnlNK1QMfAH5nrnofsBPYAKwGHtZaL0rl2EIIIYQ4ktVqhWBXxi5S6MoYpLW+MOTvNcCMkMdzgDlRtvECv0n0WEIIIYRITLa7MgohhBAiz2W1K6MQQggh8p/lKZuVUt8E/gC8z9xOA3drrR/JUGxCCCGEyAFL1QpKqcuAx4C3MUYl/AFGW4O7lVI/y1x4QgghhMg2qyUH/wtcprV+PGTZS0qpzcBVwINpj0wIIYQQOWG1QeJEYHmU5SuA49MXjhBCCCFyzWpysA74cZTlFwJb0haNEEIIIXLOarXCH4C3lFJfAFaay07DGKTonEwEJoQQQojcsFRyoLVeAXwIo2phOkZVwtuAklEKhRBCiOHFcldGc16F34UuU0pVKaU+pLVem/bIhBBCCJETVtscxPJJjHkOhBBCCDFMpJocCCGEEGKYkeRACCGEEGEkORBCCCFEmJgNEpVSZ1jYfkb8VYQQQghRSIbqrTDf4j4C6QhECCGEEPkhZnKgtZYqByGEEGIEkgRACCGEEGEkORBCCCFEGEkOhBBCCBFGkgMhhBBChJHkQAghhBBhJDkQQgghRBhJDoQQQggRRpIDIYQQQoSR5EAIIYQQYSQ5EEIIIUQYSQ6EEEIIEUaSAyGEEEKEkeRACCGEEGEkORBCCCFEGEkOhBBCCBFGkgMhhBBChJHkQAghhBBhJDkQQgghRJiibB5MKXUOcCNQCmwEfqK17g55/kfAFSGbjAKOBY7VWh9USrUBjSHP36K1npX5yIUQQoiRI2vJgVJqAvAI8Emt9Xal1M3ATcBlwXW01o8Dj5vrFwOLgZvMxEABh7TWM7IVsxBCCDESZbPk4EvAaq31dvPxfcAGpdQvtNaBKOv/EWjRWj9gPv4E4FNKLcEoUXgOuF5r7ct04EIIIcRIks3k4DhgX8jjRqAGqAa6Q1dUSo0HfgecGrK4CJgPXAkUA7PN7e7IXMhCCCHEyJPN5MAORCshiHbn/3PgZa31ruACrfWDoSsopW4DfkWU5KC+vj61SIUQQog8lK3rWzaTgwbgtJDHxwAdWuu+KOuej3HhH6SU+iGwQWu90VxkAzzRDlRbW5t6tIN2xV9FCCGEyIL0Xt+grq4u6vJsdmV8A/iYUmqa+fgS4OXIlZRSY4ATgeURT50E/E0p5VBKlQOXA89kMF4hhBBiRMpacqC1bgEuAp5TStUDHwB+p5T6sFJqfciqJwJNWuvIUoFrgUPAJoxukMuBhzIfuRBCCDGyZHWcA631HGBOxOJDwIyQdVZjJAiR2/YDF2c0QCGEEELICIlCCCGECCfJgRBCCCHCSHIghBBCiDCSHAghhBAijCQHQgghhAgjyYEQQgghwkhyIIQQQogwkhwIIYQQIowkB0IIIYQII8mBEEIIIcJIciCEEEKIMJIcCCGEECKMJAdCCCGECCPJgRBCCCHCSHIghBBCiDCSHAghhBAijCQHQgghhAgjyYEQQgghwkhyIIQQQogwkhwIIYQQIowkB0IIIYQII8mBEEIIIcJIciCEEEKIMJIcCCGEECKMJAdCCCGECCPJgRBCCCHCSHIghBBCiDCSHAghhBAijCQHQgghxP9v78zDpKiuxv327DPMxjIDwzAwbB5mGJhh2PdFBAQVAREUcMMFccMF44ILLmjc8sUvRo0mGqMmcV8So8a4xMQkRDRuGe9HVH5KTIJLjHHBBfj9UdUz1T3V3dXd1Qt43ufhYerWrap7+966de45556rhKDCgaIoiqIoIahwoCiKoihKCCocKIqiKIoSggoHiqIoiqKEoMKBoiiKoighqHCgKIqiKEoIeel8mIjMAy4DCoGXgZXGmI/D8lwNLAY+tJOMMWaJfe5s4HCsct8OrDfG7EpT8RVFURTlG0HahAMRqQJuASYaYzaLyLeBy4HVYVknAEuNMc+FXT8XOBgYCewAHgP+CtyV6rIriqIoyjeJdJoVZgF/NsZsto+vB5aJSCCYQUQKgRHAmSLyiojcKyJ97dMLgDuNMZ8aY7ZjCRrL01h+RVEURflGkE7hoA54x3G8FSgHyhxpvYEngXXAcOCPwIO2AOF2fZ9UFlhRFEVRvomk0+cgB3DzD9gR/MMY8xYwN3gsIlcB5wH1LtcHnNc6aWtrS760iqIoipJlpOv7lk7h4G1grOO4Fvi3MebTYIKIDAeajTE/ceQLAF/Z1/d2pPfG0h50oqGhwa8yU5C7hS937PTtfoqiKIqSCJP6dfH1+wawadMm1/R0mhUeB8aJyGD7eBXwYFiencC1ItLfPj4eeNkYs9XOu0xEuti+CUcAD6S60DcdPirVj1AURVGUmAytLkrbs9ImHBhjtgFHAveISBswDDhdREaJyF/sPK8CJwEP23kWAIfY5x4G7gM2Aq8Cm4DbUl3uyuL8VD9CURRFUbKKtMY5MMY8AjwSlvwh0OLIcztWDAO36zcAG1JWQEVRFEXJUtIZ1EcjJCqKoiiKEoIKB4qiKIqyO7ArfboDFQ4URVEURQlBhYMY9CxPn3eooiiKomQDKhzEoFeFCgeKonijS0Fupoug7MkEArHz+IQKBx6oLPpmvPBNteWZLoKi7NaM6d8t00VQFF9Q4UBRFEXxjZa6ykwXQfEBFQ58pG+3kva/e+1mvgrTpCrTRVCUhLj4wKZMF6GdQBrVvtnKkF5lsTMpWY8KBz7iHBe+u7QlckYHly0c1imtJAN2yy4FaY2HpSi+sWJcv0wXQXFwzJQBmS6C4gMqHKQIrzMINxulUwORTtK4hNaVsiIVUBR/KNe+lDHyc/SzkjI0zsHuSW5OcirFyYN7uKa39q1k6ei6pO6dKXrHsdqjsiQ79rHIS7IdlcyzYnxmtAnac6BYV2z4Sn5uZnqVCgceyI/Q18PVmTkObYFX06NzY6eDRvYBQoXDG5aP5ObDR2eVXTUevrXvEM95A1kytM4YUh0zz8CqLmkoSfbz5OlTM12ErEJdDqCqrDDTRVB8QIUDD2zYpybua/p0LfaUr3tp5xdpl729hvQsY05TL7p1KYj7+XGTokEtHi3Y7jRh/8nKsSl/xrxh8fe7eDh5xqC48j+/bmantAFVpX4Vx1e6lqThnVGUPRgVDjzQp8J9oFk9fWDIcQDYp7EnADUVsYWDO4+2PjBn7TuEum4d+XfaH9R0z0JSYc7aFcc+YoFAgBtXjIyZryKF22gHtTfROHx8P3pXehP+kqGxd2rjTpw2S+LK38NFkM1WjphQn5HyDqzOTmEp3SwZlRoz6CsXzkrJfbOZTC0NVeEgCdwEgB+sGMlbl831dP2EQZaPwaqpA3n2zBnt6Tvtr7TTqTHSh/uoif1jPuelC5J/oRZ7+Gi6Eanc+w3vmBWvm9cAWMKVl1UTZ8VhqoiXcQO6x8xTmiZnt10uP153D1qki+YPZX5L71QUabchLzeHQ8Yk/oEK9sl4GVHXNaHrvK5u2l1IdsXCmHr3YFLFkWy8MWiO8YG9bOEwhsYpjBfmpefzWZyhlWQqHPhMIBCIuFIh1hLFkf2sgWVRq/UhdqrZnTNwp6/D+fs3suXyeVHvW1Gcz8pJ7kJEQa63LjCib2KDntNJc+M5e7vmaffVCMSnafCK3yFt57fUJnxta9/kZgGbztsnZp5Zjb347tIRDKutiJhHI/n5S7Tf2gv7NqXWhJR+/HuPnQ55ebk5PHLy5LjvEcvJeOnourhn6HevGh93ORLBbZKQDlQ48Mhp++wVM0+sfRhiBRrq07WELZfPa1+ZsLC1Y7YedNYrzs9NyDnxvP0aXdN7V1plHtIzeuASp9nDK/etnkCJQ+qtjhAYKtj1czzaURpr4pPwna9Wj9LoM28vL2IyvhF5HoWxePO68eAJEyOeu+u4xAa2hSM6BKN0BbuZPdQy1T10YuT6OJnuQ0CveD3ua9rf/QyvB84SUvk9S8TcFnxle1cUce0hIzqfDwTibrkuhXv2clkVDjwSSy01Y0g1/+vS6RKhe2khmy/dl6Mm1renFeTlcMPyVp5ZOw2A8/drDFFFJrrUcWS/bty/egKrpw/qZMt3Li0cHUHNF4nz9muk1aO2ocx+ybzOvprrKnny9KkJRXUc2z/UbDBlL+seGxYM46CRfdi/2T91/KLWzqaYeJYlBds03qBYQRkrx4MUE6tfh3Pl4ub2fnLz4aMAb3E5SvI7l2XhiNqI2iQnJ80YzKvrZzO8T2UnVe7mS/eNeF0yviklBbk8d9YMbjlitKf8mVql8P1lrZl5cBjh5fBTNvBD0Jg1tCdtF83hqbXT2H/4nqalSQ0qHCRJ8MO2f3MNlT56SOfn5nQyT8xpqmmffR81qX+IevuQMX0939u5kVQgYJkMcnMCXHvIiJDZ4C9PnswPDx/FlsvnURSnra+hxrpPsAZ728sDu7rEMpjZ2JP7Vk9wjRYZiQFVpdx65Ji4ygTQPUxz0NfWiNT3KOGqxc2u9ZQYWhU3fnP6VK4+uDkk7arFzeS7aAPKIsxA3PK64TYTShW5OQFKw8p7Qphjrht3HNw57sCQmrKI2iQnOYHOzwwS7Tc6YkJ9zHtHIkCA3pXFntsgGqlcpz42zDy08dzYwpYb/brHF3jtmMmhZsq5KV5ZEw0vfjij67tRXJBLYV5uRLNvMq10wf7umlnw158knXopFQ7iJDxQUdDxpqp099hLIaiiDeKcXFaVFYY4ONZWFrN3Q0f+BSNquXHFyLjV+m4EX9AL92+kW5cCWvt2jVsA8YN18xr5nyUtjPfgiBjOvk29YuY52+E8WV1W6DoL2njuTO5fPaH9+KXzZ3HX0s4f00h206owr3xnrk3rZnL14lAhJVkmDrJ+q0gfbDeKXJy3ojnTzh3W8dv2cZi04pmhezHLDO8Tn69ALLMUdP7QnjDdWjJ6UpxLRxOhuqxjHPIqHN17/HjuO35C7Iw2OYGOOjm5aP5QfnSEpU2q62r9BoN9WL0RSzPl1+qlRLU/uTmBqKuckvFRgszt16HCQQJcuqCJcQMsif2E6YO44+ixTIoQ3RCs5TfZsATn9Yvn8P1lI1k5qmO2EU/goe8saWH20OgfxOlSRdeSfJqSdNCKhJcolOGzKSfhH+ei/FwOHFEb8gIeEMPTP3gPL5qi46YOZNO6mZwwfSCTBrn3keKC3PYP7aDqUipK8ikrDBWUXr5wFn/xuOqk0CFkdS8tpLVfYs6kkbjkwGE8fca09vp79RX5oW2GCOLl433doa2UF3UM/ol6q4fPdL1Sa8crGV0f+hsO7V3O8uaONGe/irQcN92D/IUHDPWUb2S/bq7xVuLlsPH1zBhiTSaKC3LZcvk8z9rAZ8+cHprg+KmuWZLYzHtYzw5Byctvn4j54onTpvCHs2ck1bZnzIrtz5YJVDjwSK7d+IV5OSwb24+fHWs5dOXmBJgYYdD/1SmTufjAJsqK8ikrymflpAEps032CItKdr6LA2JRfi65OQFmDizjFydNAmBqmN3ey2qBqxY3M6uxp+u5xaPqePH8WSEDuhuT7Nnn8Ag270mDevD7s2Z0Sq/xEI750gWRByQv9dtveIdwcMregyPm89qW3UsLWTt7CDk5kZ2eYpWqvCjf00z9sTVTfJlJHTo2spmqIC+H+h4dESK9zozEgwPjgBiRJxN10jxnbmJLE/v36MIfzp7BqqlB04nV6IEATKp3K2vAc/jt5ji1FsnwGx8jWf70mHEhjsbRiJQv3HekLop2YHiMiUakVQbrpnWMUany+B9UXUZ1WdFuFcDNKyoceGTCwO6snjaQyxYO93xNQ015yLLDkf268tZl0ZcdJkptZTFPnzGt/XjCoOhq8qbaCl5bP7uTrdCLar+xdzk/OGxUp/RDxtQxs8FdaAh/Nfdp7MVr62d7dlqMj8gDQbxjxKkeVqkEtUhOvC4RDcdtjIk17gyt7TDzuKlxo11fnO9eznjKX+Djeu/H1kzBXDKn/TiRpa1u2qVkZnY1FcVxCfWJPCvR4uV53OSotyMmS3VZYcKmpjUzBzN2QHfPbd7Yu5zrDu3sNPni+R1LctfMjCyAQ2zH2g0Lh4WY74KUF3nTMvkRgTZcCOqkCUkCXcqY5eTkBDhzzpCsjhvunM156fBuS3Gcs+Z4uWzh8E6DRrRBz+35tXbkwfEDu7d/1Goqitq1CPHYud1IxWsW/M3G9u/GX87fh2sPGRF1JuSVkoJcVk0dyN2rQu3BVywazsMnTmo/jqWlica1S1PrzPidJc1xBRTKz82xnMYSdA87cmI9G2Kosl9wxIqI1E5Darw7oQ4KE8gG9Ehs343jpsYXOKh/jy5UlOR7WrHgfA8PaO7NIhcb+YENob5EfbuVdBK0EmmXecNr+O3a6dx21BgWtfbh+mWtIR/TSH4fVywazj0eYgkU5ee2mzF7lBZQWZIf0ck3nHtWjefRU6y4CfXd/dsvpa5bCTeuGOlqMvBzRVQq2bMXan6DcTomxUNuToCe5YX86+MvfC6RRayP84CqUp47awa9yov458fb29NrK4tZN68hqld0Y005dx4zlvc/+TIk3TmcOYXwRHeBDA60znu9ucGKipmTE+CAKC9/+CxgYhQNTyAQcI0GeXCUZavxzkC9rBZIhmG1FQyqLqOtrS3k97o3Dge4eLhg/9h2dqfgHP5z9elazKNrpnQSQhtsJ9yDRvbhhmfeiHr/cM1B8KNTH2NFwBmzhBufeTNqHifBKKNzh9WweGSfuFcchHPcmB480PZx+3EgACdOH8R3f7M5qfsC9O1eQt/uJe1Lh70QrZ9HYq+eZdx5zDgA2tramC5VPGXei7jqZJRjifbycf249JG2uJ8ZidlDe7n6aEXS1mUbu0cplbTy4AmTPK/vTpRo37DelcWuqsSjJw/otKeB00ZdWphnO8mFfoBD7dTWuYFVXXhgtbegOrE+uIGAJRR4iSsQzo/t5ZgZ0hxmhNrK4vZooPHSP8FZOVh27kiRQoP86pTJrtqpmopitlw+L+7NsG46bBTzW3pz96rxLBhRy2Hj+7ULCeFNHmvZZDS/gSsXN3PijOjq+UQI7/vO45fOT5+T9d2rxnO6w8R30Xx3ITD8Pbrm4BYunj+0U2jk6VLFTWGm0VREZ/XKiCQjp6YCFQ4yhNt6/2yhV0UR0z1sW+yFwdWWenZOjFUOiXL3ceNZOzv6BkLVDlNQcPA4ZEzfEDNMpgh3sItn5v/9Za3t8SOiObeFO6vuDgQdIsN9Un6wYmTcyw+DmEv2jRgpNEhZEiYaN/Zp7EkgEGB0fTcCgQAXzW/if6KYchaMiOzcOdCxA+aUvaoSjuMQ7GMbz9mbTWE7bcYy2zm7Z4VPY5iX33x0fTdOcjgHh2tGI702XbsUsGJ8fSdtzi1HjmnfJC/IzjDZ4OU4V5hFi0gajYNH13Hn0eP4kx0QLFo02lTGzAhHzQoZ4Jm105KyE0fjFydN4ssdOwF45OTJcUfXS5TaCLsU9u1egrlkDoV5iZcj2uvQvbSQUXHMQmc39eJnf34n4gqT3Ym5w2razSy3HTWWN97/xNUZrrQwjy2Xz+Mf//mcdz/6POISzIFVXXjjvU8BGFXflVuf25KScnsRgCYO6uG6Z0hlSQHTpJqXt/6n07me5fELQc5lmF78dIKmKGuDpR1xP8+J28/gdfC/7aj4A4AFCQrIbialX582hfGXPQlY5VvU2ocfP7eFSYOrePildxN+ZjTieX878H+W71T8PX7qlLjH6Oa6SnqWF8YVkA46BJ3iglx+c/pUenQppPmix0PyVJbk89FnXzG+Ln0TGhUOMkA/Hx1fwnHGF0j1lr9OosW9T1QwqKko4pjJ/VmSYGhoN8bUd4u5UdXuSEVJfsyVHzUVxVG3El8+rh/rH/4rkPhGQjMbevJE279oqi3n1b9/nJod5VxsMI+tmZKQs3DwgzBlryquWBR7JVJdtxIeXTOZAT1K2fSK9VvNauzJ37Z9Evez04nX5ZXh/aOuWwkvnj+Lm599k4dfejdhPx2/SYUZrqQgj1uOGE1zXaUnQdEtMuOfzpnpktM7Tu2Qk+8saeG7T2yma3H6AsWpcKD4gtcgKl1LCvjos688Bc4JBAKcOy+6Gjgelo3t63mzlGsObm43iaSDTNo7gwSAJ06z7NpOAfYXJ01i23+3R7gqlBuWt/LVjl3s3LWLF9/+KESjFNzMKN6tcb3gJYZCNA5s6R1z47QgQ3pZ5a8syuWF8/ahsjifVbdvSur5QVrqunLX81t9uZeTROJDOGfAR0yop7Qwj8Wj/BPUkyFc61Fux/ZI1ikzkjn1psNGccxtz7cf//Wi2Z5jPYAVzOveF7by5OvbPOU/ZnJ/frrxHT754murXFLNdKmmrc0/h8lYqHCgpJXbjx7Lk69v881eGQ+Hx2GjXeiyaVI6SHQJnx/sovOyPAhqo6JrEq47tJW6bsXk5eYQVBSFRw3tUVrIvcePb/f8T6acyfLM2mkU5uXy7UdfT+o+wRnmVKni8b/+i4ExgjjF4pAxddRUFHHkrX9O6j5+cNzUjj0z8nJzWBqnujyVhDu0NtVW8KMjRjFhYGrMheFLtOMRDMBazvnJF1+1Cwe3HjmaI26J3Mbnzmtkn8ZeHHzjH3yJw5AIKhwoCfPM2mlMvfLpuK6prSwOCQzlB0EP9gOjOHNlC0F16PXLWhmRkgBQmWGex53uRvaLb3dPPwjfDwU6NCPBuP09kgwffOiYvuzbVJP0QB4IBJg+pNrahfLCxzo5yfnBrBjOwfesGp/07xEPfoWVDoZu9pNVUwfGXIKaCNMktsP36PquXLqgKerS6FSiwoGSMKn0nYiH6vIiHjmsPw0N7irPyYOr2LztE9/tpXOaevHTjW8zKoEPXmlRnmc1dqa4eP7Q9r0FsgmvezmAtfFUaVHkYe6kGYNo6VsZ1/p7NwKBgK8zvNLCPB46cRK//uu/OsUZSNZjfUyUvUcgdO3/N51gnJFn/u+9lNz7nQ8/i3g+EAiwbKy/E6l4UOFA2SMIBAIRZyDnzB3CUZPqEw4MFeS4KQO48bdvtjttTd2ryjfnxqBtPrjLZzawYnx9Wp8XHm8/EsdOGcAHn37B2fvGjrwYyxcmLzeH6R5mcX7iFg7XLcZBU20FTbUVIcLBr0+dkhGTXLpZ1BpbC7iotQ87du5MQ2n8DWEcvNWqqbG3Os8kKhwoezx5uTn06Zq8anB+Sy1nJ7iBT5B1+zVw7v2vdtI2lBXlZ3wVRaYDMcWKVxGkS2Eelxzobbe/rCYODUgwjsXgnvE7XjY7NiZK1icilfzx7L354usdnjWSVx/s71bkqWZ+Sy3Pb/m3a9TTbCStwoGIzAMuAwqBl4GVxpiPw/IsB9Zi+R19BpxsjHnePrcJKAaC8XHvMMZcmabiK0rSDOlVnrLQwbs7R0/OHq1JslQU53vWhMTid9+aTlcP24O78fy6F5PhLAAADo5JREFUme2Bjd7YMDeD7q6dOXRsX+7809vtx9luZkuWovxcrkxww6tMkDbhQESqgFuAicaYzSLybeByYLUjjwBXAq3GmH+IyFzgPqCviHQBBgJVxpiv0lVuRVFSS35uIGVe5pniRcfmTtEY078bG9/6MGqeZLReTsdCt90qM8kl85tct5bPRjK/0Dj9pFNzMAv4szEmaEC7HnhJRE4wxgR/+y+Ao40x/7CPnwd6iUgBMAb4BHhURKqBJ4BzjDGfp68Kyu5AT583EwrOvLJtcPWLvJwAX6fCLd4jmy+dm7Fnpwqv+2z89Jhx7Mjgb59JcnICFOWkL6iPEh/p3FuhDnjHcbwVKAfajWjGmC3GmF8CiEgAuAZ4yBjzpZ3vKWAxMBroi2WiUDLI7SvHxtyPPd34bb+/blkrZ84R9urpHr1sd+fiA5sAqO/h/5ItJTq5OYFOa+iTpUdpYcL7TyhKkHRqDnJw1850ClBumxBuxRIo5gAYYx4CHnLk2YBlclgTfr3fUaS2b9+e1shUqcbP+nQHZtf6/5vHS7Q6+VG26T3h9deTC5gTD+nsc82lu/juvFpqdn1IW1t0FXei7GnvECRepy3vWdEmt3/+eUp+k58ssjz94713Mm2UjW3rZ5/7YFtohNBM1Ted71E6hYO3gbGO41rg38aYT52ZRKQv8DDQBkwPmg1EZH/gP8aY39pZA4Cr70FDQ3Ie5eG0tbX5fs9MsqfVByLV6U3A//6QDtLdRqm2/H5z+lxsutduh0feZeHo/jQ0ZM9ytkTqs7D1C2Y19qShIb6trNOBn32uoQHyK6r56LOv6NqlgIaGzAQmSsV7tGmTe+jvdAoHjwNXi8hg2+9gFfCgM4OIlAFPAz82xqwPu74PcJ6ITMVarXAa8POUl1rZbelSkMunXya3c56i+E11eRGvrZ+dth1TU8k1B7dkughpY35L9kdg9ZO0CQfGmG0iciRwj+1g+AZwmIiMAm42xrQAJwL9gAUissBx+d7AjcAA4AW73E8BF6Wr/Mrux8ZzZ7Ij04v3FcUFrxuAKUqmSGsPNcY8AjwSlvwh0GKfv4zoToZr7X+KEhMdgBVFURIjnasVFEVRFEXZDVDhQFEURVGUEFQ4UBRFURQlBBUOFEVRFEUJQYUDRVEURVFCUOFAURRFUZQQVDhQFEVRFCUEFQ4URVEURQlBhQNFURRFUUJQ4UBRFEVRlBBUOFAURVEUJQQVDhRFURRFCSGwaw/btW7Tpk17VoUURVEUJYWMHDkyEJ62xwkHiqIoiqIkh5oVFEVRFEUJQYUDRVEURVFCyMt0AbIZEZkHXAYUAi8DK40xH2e2VJ0RkeXAWmAX8BlwsjHmeRHZBBQDX9pZ7zDGXCkiJcDNwAgsAfFbxpgH7HuNBb4HlALvAsuNMf9Ia4WsclwNLAY+tJOMMWaJiJwNHI7Vd28H1htjdolIFXAb0A/YCRxrjHnOvldG21FEDgNOcyRVAH3sf68BWx3nrjTG3JGt9RGRAHAr8Iox5ioRyQWuBuZgtclVxpgb7LyDgR8CPYBPgMOMMa/b544CzgDygSew+uxX0fpmmupTDFwHjAECwJ+AE4wxn4vIMOAPwN8ct1hijDHR3ptIfTZddbLT3senfhatzdNRHxG5BxjkyNIfeMYYc4CI7A/8GHjbcX6yMea/2VCfKGO1b+OaX/VRzUEE7Ea5BVhkjBHgTeDyzJaqMyIiwJXAHGNMC3AJcJ+IdAEGAs3GmBb735X2ZRcCnxhjGoB9gOtEpI+IFAD3AGvsc/dgDe6ZYAKw1FH2JSIyFzgYGAk0AdOxBAiwBvRnjTGNwHLgbhEpyYZ2NMbcFqwHMBr4J3AiUAl86KhjizHmjmytj4g0AL8BDnIkHwfshdUeo4E1IjLGPncHcINdhwuAe0QkICJNwHpgKiBYv8Op9jUX4tI301ifc7EG1OH2v2LgbPvcBODOsPYy0d6bGH02LXWyxwg/+1m0Nk95fYwxBznep2OAj4AT7NMTsD6Gzrr+NxvqE2Ws9ntc86U+KhxEZhbwZ2PMZvv4emCZLcVmE18ARztm988DvYBJWLO1R0XkFRH5jj0rAlgA3ARgjHkb+DVW5xwNfGyM+b2d74fA3iLSPT1VsRCRQqyZ45l22e8Vkb52ue80xnxqjNmO9XIsF5E8YD9Hnf4CbMaSnLOtHb8FbDPG3Ig1kO0QkWdF5GUROV9EcrO4PidgzervdqQtAG4xxnxtjPk38DOsNqkFhtjHGGN+hTWrHgHMBx4yxrxnjNkJ3Ig18AXv59Y301Wf3wKXGGN2GmN2AC9izdrAaq8GEXlBRDaKyEI7Pdp749pnU1SfSHXyu5+5tnka6wOALZT9GEsoe8dR1xki8pJd3yl2ejbUJ9JYvRh/xzVf6qNmhcjUAe84jrcC5UAZkDWmBWPMFmALtKvfrgEewlI1PQWcgqW+ugNLBbUG97r1wVKHtqcbY74UkfeAWuCD1NYkhN7Ak8A6LLX7GcCDwDasWUSQYLl7ADnGmPdczpWQJe0oIj2A07FmCGC9f08AZ2Gp139pl+lnZGF9jDEnAojILEeyW18abqe/a3/8nef62Oe2uKRHul9KNAdu9THGPB78W0T6Yb0vx9pJnwI/xRJm9gKeEZG3w8sc9t7U4d5nU0KENvK7n0Vqc9+JUJ8gK7H62P2OtA+AO4F7gYnAgyLSHKHMaa1PlLG6Bngs7PnJjGu+1EeFg8jkYNmFwtmR7oJ4wTYj3IrVMeYYYz7C6njB8xuA+7AGu/C6BbDq5Vbn4Lm0YYx5C5gbPBaRq4DzgPeJv9zZ1I7HAg8aY94EMMbc5DwpItcAJwN3sXvUBxLrS5GuiXa/tCIiI4H7ge8ZY34BYIxZ7cjSJiI/B/YHDInVNS2koJ9lvE42p9IhuAFgjFnoOPydiDyHZZ7KmvqEj9V0bodkxzVf6qNmhci8jTWDDVIL/NsY82mGyhMRW+X+HFYHmG6M+UhE9neo1MDqIF/Zf4fXrTeWdBmSLiL5QHfg7yksfidEZLiIrAhLDgD/D/dybwMCItLN5Vw2teMSLJUhACKyQkScEn2wjXaX+uBSHmc5a8LMHZHqEEyPdr+0ISJLscwZZxljNthpuSJyroiUObIG2yvae5MN9fG7n2VDnUZgTW6fcaRVisg5YX3OtY3IUH3cxuooz89o+6hwEJnHgXFieVwDrMJSbWcV9mD1NHCfMWapMeZz+1Qf4CoRKba9V08Dfm6fexBb4radveYAv8DyzO4uIhPsfEcBf7A7cDrZCVwrIv3t4+OxvHEfxLKrdbH9Eo4AHjDGfI2lKg3WaTjQiPW7ZEU7ikhXLA/r5xzJTcBF9oenGMtJ8ee7Q30cPAgcJSJ5IlIJLMVqk61YXv1LAERkNla7voKl0TpARKrtgfxY4AHH/dz6ZloQy9v9WmCWMebOYLrtf3CAo2z9gEVY6uto741rn01TdYL43c9c2zxdlbGZCjxpQld9/BfLR2EhtAsQY4BHyYL6RBmr/R7XfKmPmhUiYIzZJiJHYnlYFwBvAIdluFhunIjlMLVARBY40vcGBgAvYLXzU8BF9rkLgOtF5DUgF1hrjHkDwHay+p6t+vqADNTZGPOqiJwEPGwLNluBQ4wxb4u1nGwjUID1EtxmX7YauFlEXsVSqa0wxvwHIEvacRDwD2PMV4609VjL317BsgXfjeV8BdlfnyDXY62KeQmrTW40xgRnc4cAN4nIOmA7sNj2QXhZRC7C8ivJx/q4ftu+JmLfTBNXYc02bxaRYNrvjTEnAMuAG0TkCLtsa4wxbRD5vTHGPBylz6YLv/tZtDZPF4MJ9VvBGLNDROYD/ysi64GvsZaavg9ZUZ9oY/V9+Deu+VIfDZ+sKIqiKEoIalZQFEVRFCUEFQ4URVEURQlBhQNFURRFUUJQ4UBRFEVRlBBUOFAURVEUJQQVDhRFAUBEnhaRS+y/DxKRXil8Vvv9ReRCEfldqp6lKEr8qHCgKEoIdoCfu7E2SkrH/a/CCjCkKEqWoEGQFEUJJ9U7PIbc3xjzSYqfpyhKnKhwoChKOG/Z/28WkSONMbeKyIHApUB/4HXgHGPMo2CZI4BXsUIdFwPNWBHsrsDagXIX8Cyw0hjz9/D7A/XATGPMJPt+47H2vR8BvAdcaYy5zj53K/AfoBpL2/B3YIMx5tZU/BCK8k1FzQqKooQzxv5/PPBzsba8/QlwOTAM+AFwv4i0OK45Eism/Hysfet/ibVN8FCsvecHAOe63d/5YBFpwAqr/Fss4eAC4AoRWezIdjzwItCKtdXt9WGb0yiKkiSqOVAUJZzg/vHvG2M+F5EzgB8ZY35ip78hImOBk4CVdtqvjDG/A7AdDTcAV9sb47wlIvcCEyLc3/nsY4CXjTHn2Mf/ZwsMZ2L5KQC8Yoy5wn7WuVgx65uwBApFUXxAhQNFUWLRAAwTkZWOtHysjWKCbAn+YYz5p63+P9XWLjRimRr+5PFZ4fmew9ptL0j7RkzGmI9t4SLfw70VRfGICgeKosQiD2tFwS1h6V84/t4e/ENEaoHnsVT/jwE3AfOASR6etd0lLZfQsepLlzypdqJUlG8UKhwoihJO+FatBhhgjPlbMMHeEvcD4FqX6xcAHxtj5jryn0THBzzaVrBtWFvYOhlvl0FRlDShwoGiKOEElxY2i8g/ge8AvxORjcDDwEws58L5Ea7/AKgVkX2wTACLgUVYmgS3+zv5PrBGRDYAtwLjsEwKpyRbKUVRvKOrFRRFCcEY8wHWh/lOrOWHfwSWYTkLvgacChxpjPllhFvchbW64S5gE5Ym4FRARKQ4/P5hz96KZYKYDbwCnAecboy52ccqKooSg8CuXdE0fIqiKIqifNNQzYGiKIqiKCGocKAoiqIoSggqHCiKoiiKEoIKB4qiKIqihKDCgaIoiqIoIahwoCiKoihKCCocKIqiKIoSggoHiqIoiqKEoMKBoiiKoigh/H+UsosA1ivZ7QAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 594x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "thunderstorm_predictors = data.drop(['DAMAGE_PROPERTY_NUM',\n",
    "                                     'BEGIN_LOCATION',\n",
    "                                     'END_LOCATION',\n",
    "                                     'CZ_NAME_STR',\n",
    "                                     'BEGIN_DATE',\n",
    "                                     'END_LAT'], axis=1)\n",
    "add_mean_income(thunderstorm_predictors)\n",
    "(predictors_Tensor_train, predictors_Tensor_test, outcome_Tensor_train,\\\n",
    "outcome_Tensor_test, outcome_mean, outcome_std) = preprocess(thunderstorm_outcomes, thunderstorm_predictors)\n",
    "\n",
    "without_end_lat_model = leave_out_Net()\n",
    "\n",
    "# Train one layer\n",
    "train_neural_net(without_end_lat_model, outcome_Tensor_train, predictors_Tensor_train,\n",
    "            outcome_Tensor_test, predictors_Tensor_test)\n",
    "\n",
    "one_layer_mse = compute_MSE(predictors_Tensor_test, outcome_Tensor_test, without_end_lat_model).item()\n",
    "print(\"Difference with original mse\", one_layer_mse - one_hidden_mse)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>Leave out END_LON</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 606,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 8)\n",
      "Shape of inputs array test (71, 8)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "1.0689170360565186"
      ]
     },
     "execution_count": 606,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 594x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "thunderstorm_predictors = data.drop(['DAMAGE_PROPERTY_NUM',\n",
    "                                     'BEGIN_LOCATION',\n",
    "                                     'END_LOCATION',\n",
    "                                     'CZ_NAME_STR',\n",
    "                                     'BEGIN_DATE',\n",
    "                                     'END_LON'], axis=1)\n",
    "add_mean_income(thunderstorm_predictors)\n",
    "(predictors_Tensor_train, predictors_Tensor_test, outcome_Tensor_train,\\\n",
    "outcome_Tensor_test, outcome_mean, outcome_std) = preprocess(thunderstorm_outcomes, thunderstorm_predictors)\n",
    "\n",
    "without_end_lon_model = leave_out_Net()\n",
    "\n",
    "# Train one layer\n",
    "train_neural_net(without_end_lon_model, outcome_Tensor_train, predictors_Tensor_train,\n",
    "            outcome_Tensor_test, predictors_Tensor_test)\n",
    "\n",
    "one_layer_mse = compute_MSE(predictors_Tensor_test, outcome_Tensor_test, without_end_lon_model).item()\n",
    "print(\"Difference with original mse\", one_layer_mse - one_hidden_mse)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>Leave out MEAN_INCOME</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 607,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 8)\n",
      "Shape of inputs array test (71, 8)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.9976235032081604"
      ]
     },
     "execution_count": 607,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 594x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "thunderstorm_predictors = data.drop(['DAMAGE_PROPERTY_NUM',\n",
    "                                     'BEGIN_LOCATION',\n",
    "                                     'END_LOCATION',\n",
    "                                     'CZ_NAME_STR',\n",
    "                                     'BEGIN_DATE'], axis=1)\n",
    "# add_mean_income(thunderstorm_predictors)\n",
    "(predictors_Tensor_train, predictors_Tensor_test, outcome_Tensor_train,\\\n",
    "outcome_Tensor_test, outcome_mean, outcome_std) = preprocess(thunderstorm_outcomes, thunderstorm_predictors)\n",
    "\n",
    "without_income_model = leave_out_Net()\n",
    "\n",
    "# Train one layer\n",
    "train_neural_net(without_income_model, outcome_Tensor_train, predictors_Tensor_train,\n",
    "            outcome_Tensor_test, predictors_Tensor_test)\n",
    "\n",
    "one_layer_mse = compute_MSE(predictors_Tensor_test, outcome_Tensor_test, without_income_model).item()\n",
    "print(\"Difference with original mse\", one_layer_mse - one_hidden_mse)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Since the MSE is only changing very slightly, we dare not make any hard conclusions about exactly which features are highly impacting the predictions of our model. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2>Finding the best model over 100 iterations</h2>\n",
    "<p>It seems that whenever we retrain the model, the MSE is slightly different. Therefore, we decided to retrain the one hidden layer and the two hidden layer neural net 100 times and seeing what the best MSE is we can achieve. </p> "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 9)\n",
      "Shape of inputs array test (71, 9)\n",
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 9)\n",
      "Shape of inputs array test (71, 9)\n",
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 9)\n",
      "Shape of inputs array test (71, 9)\n",
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 9)\n",
      "Shape of inputs array test (71, 9)\n",
      "(281, 1)\n",
      "Shape of outcome array training (281, 1)\n",
      "Shape of outcome array test (71, 1)\n",
      "Shape of inputs array training (281, 9)\n",
      "Shape of inputs array test (71, 9)\n"
     ]
    }
   ],
   "source": [
    "# *********************\n",
    "# NOTE: the code below takes a while to run, as it as to go\n",
    "#       through 100 iterations of retraining, it may take up to \n",
    "#       about half an hour for this cell to complete\n",
    "# *********************\n",
    "\n",
    "best_model = None\n",
    "best_mse = 10\n",
    "for i in range(100):\n",
    "    data = data.sample(frac=1)\n",
    "    thunderstorm_outcomes = data.iloc[:, [6]]\n",
    "    thunderstorm_predictors = data.drop(['DAMAGE_PROPERTY_NUM',\n",
    "                                     'BEGIN_LOCATION',\n",
    "                                     'END_LOCATION',\n",
    "                                     'CZ_NAME_STR',\n",
    "                                     'BEGIN_DATE'], axis=1)\n",
    "    add_mean_income(thunderstorm_predictors)\n",
    "    (predictors_Tensor_train, predictors_Tensor_test, outcome_Tensor_train,\\\n",
    "outcome_Tensor_test, outcome_mean, outcome_std) = preprocess(thunderstorm_outcomes, thunderstorm_predictors)\n",
    "    \n",
    "    # define our nets\n",
    "    one_hidden_model = one_hidden_Net()\n",
    "    two_hidden_model = two_hidden_Net()\n",
    "    \n",
    "    # Train one layer\n",
    "    train_neural_net(one_hidden_model, outcome_Tensor_train, predictors_Tensor_train,\n",
    "                outcome_Tensor_test, predictors_Tensor_test)\n",
    "    # Train two layer\n",
    "    train_neural_net(two_hidden_model, outcome_Tensor_train, predictors_Tensor_train,\n",
    "                outcome_Tensor_test, predictors_Tensor_test)\n",
    "    \n",
    "    one_layer_mse = compute_MSE(predictors_Tensor_test, outcome_Tensor_test, one_hidden_model).item()\n",
    "    two_layer_mse = compute_MSE(predictors_Tensor_test, outcome_Tensor_test, two_hidden_model).item()\n",
    "    \n",
    "    if (one_layer_mse < best_mse):\n",
    "        best_model = one_hidden_model\n",
    "        best_mse = one_layer_mse\n",
    "    if (two_layer_mse < best_mse):\n",
    "        best_model = two_hidden_model\n",
    "        best_mse = two_layer_mse\n",
    "        \n",
    "print(best_model)\n",
    "print(best_mse)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From training the one hidden layer model and two hidden layer model 100 times on random training and test splits of the data, the best MSE we can achieve is around 0.535 (note that this result may differ when rerunning the cell above because both the data and the optimizing algorithm have random elements in it). We can see from the results above that the two hidden layer model is more prone to overfitting. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 610,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "one_hidden_Net(\n",
      "  (fc1): Linear(in_features=9, out_features=6, bias=True)\n",
      "  (fc2): Linear(in_features=6, out_features=1, bias=True)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "print(best_model)"
   ]
  }
 ],
 "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": "3.7.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}