neural_net.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.12.0\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "np.set_printoptions(precision=3, suppress=True)\n",
    "\n",
    "import pandas as pd\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "\n",
    "print(tf.__version__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>CensusTract</th>\n",
       "      <th>State</th>\n",
       "      <th>County</th>\n",
       "      <th>TotalPop</th>\n",
       "      <th>Men</th>\n",
       "      <th>Women</th>\n",
       "      <th>Hispanic</th>\n",
       "      <th>White</th>\n",
       "      <th>Black</th>\n",
       "      <th>Native</th>\n",
       "      <th>...</th>\n",
       "      <th>Walk</th>\n",
       "      <th>OtherTransp</th>\n",
       "      <th>WorkAtHome</th>\n",
       "      <th>MeanCommute</th>\n",
       "      <th>Employed</th>\n",
       "      <th>PrivateWork</th>\n",
       "      <th>PublicWork</th>\n",
       "      <th>SelfEmployed</th>\n",
       "      <th>FamilyWork</th>\n",
       "      <th>Unemployment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1001020100</td>\n",
       "      <td>Alabama</td>\n",
       "      <td>Autauga</td>\n",
       "      <td>1948</td>\n",
       "      <td>940</td>\n",
       "      <td>1008</td>\n",
       "      <td>0.9</td>\n",
       "      <td>87.4</td>\n",
       "      <td>7.7</td>\n",
       "      <td>0.3</td>\n",
       "      <td>...</td>\n",
       "      <td>0.5</td>\n",
       "      <td>2.3</td>\n",
       "      <td>2.1</td>\n",
       "      <td>25.0</td>\n",
       "      <td>943</td>\n",
       "      <td>77.1</td>\n",
       "      <td>18.3</td>\n",
       "      <td>4.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>5.4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1001020200</td>\n",
       "      <td>Alabama</td>\n",
       "      <td>Autauga</td>\n",
       "      <td>2156</td>\n",
       "      <td>1059</td>\n",
       "      <td>1097</td>\n",
       "      <td>0.8</td>\n",
       "      <td>40.4</td>\n",
       "      <td>53.3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>23.4</td>\n",
       "      <td>753</td>\n",
       "      <td>77.0</td>\n",
       "      <td>16.9</td>\n",
       "      <td>6.1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>13.3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1001020300</td>\n",
       "      <td>Alabama</td>\n",
       "      <td>Autauga</td>\n",
       "      <td>2968</td>\n",
       "      <td>1364</td>\n",
       "      <td>1604</td>\n",
       "      <td>0.0</td>\n",
       "      <td>74.5</td>\n",
       "      <td>18.6</td>\n",
       "      <td>0.5</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.5</td>\n",
       "      <td>19.6</td>\n",
       "      <td>1373</td>\n",
       "      <td>64.1</td>\n",
       "      <td>23.6</td>\n",
       "      <td>12.3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>6.2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1001020400</td>\n",
       "      <td>Alabama</td>\n",
       "      <td>Autauga</td>\n",
       "      <td>4423</td>\n",
       "      <td>2172</td>\n",
       "      <td>2251</td>\n",
       "      <td>10.5</td>\n",
       "      <td>82.8</td>\n",
       "      <td>3.7</td>\n",
       "      <td>1.6</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>1.6</td>\n",
       "      <td>25.3</td>\n",
       "      <td>1782</td>\n",
       "      <td>75.7</td>\n",
       "      <td>21.2</td>\n",
       "      <td>3.1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>10.8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1001020500</td>\n",
       "      <td>Alabama</td>\n",
       "      <td>Autauga</td>\n",
       "      <td>10763</td>\n",
       "      <td>4922</td>\n",
       "      <td>5841</td>\n",
       "      <td>0.7</td>\n",
       "      <td>68.5</td>\n",
       "      <td>24.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.9</td>\n",
       "      <td>24.8</td>\n",
       "      <td>5037</td>\n",
       "      <td>67.1</td>\n",
       "      <td>27.6</td>\n",
       "      <td>5.3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 37 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   CensusTract    State   County  TotalPop   Men  Women  Hispanic  White   \n",
       "0   1001020100  Alabama  Autauga      1948   940   1008       0.9   87.4  \\\n",
       "1   1001020200  Alabama  Autauga      2156  1059   1097       0.8   40.4   \n",
       "2   1001020300  Alabama  Autauga      2968  1364   1604       0.0   74.5   \n",
       "3   1001020400  Alabama  Autauga      4423  2172   2251      10.5   82.8   \n",
       "4   1001020500  Alabama  Autauga     10763  4922   5841       0.7   68.5   \n",
       "\n",
       "   Black  Native  ...  Walk  OtherTransp  WorkAtHome  MeanCommute  Employed   \n",
       "0    7.7     0.3  ...   0.5          2.3         2.1         25.0       943  \\\n",
       "1   53.3     0.0  ...   0.0          0.7         0.0         23.4       753   \n",
       "2   18.6     0.5  ...   0.0          0.0         2.5         19.6      1373   \n",
       "3    3.7     1.6  ...   0.0          2.6         1.6         25.3      1782   \n",
       "4   24.8     0.0  ...   0.0          0.6         0.9         24.8      5037   \n",
       "\n",
       "   PrivateWork  PublicWork  SelfEmployed  FamilyWork  Unemployment  \n",
       "0         77.1        18.3           4.6         0.0           5.4  \n",
       "1         77.0        16.9           6.1         0.0          13.3  \n",
       "2         64.1        23.6          12.3         0.0           6.2  \n",
       "3         75.7        21.2           3.1         0.0          10.8  \n",
       "4         67.1        27.6           5.3         0.0           4.2  \n",
       "\n",
       "[5 rows x 37 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('archive/acs2015_census_tract_data.csv')\n",
    "concat = False #change as needed\n",
    "if concat:\n",
    "    df = df.rename(columns={'CensusTract':'TractId'})\n",
    "    newdf = pd.read_csv('archive/acs2017_census_tract_data.csv').rename(columns={'VotingAgeCitizen':'Citizen'})\n",
    "    df = pd.concat([newdf, df])\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_loss(history):\n",
    "  plt.plot(history.history['loss'], label='loss')\n",
    "  plt.plot(history.history['val_loss'], label='val_loss')\n",
    "  # plt.ylim([0, 10])\n",
    "  plt.xlabel('Epoch')\n",
    "  plt.ylabel('Error [IncomePerCap]')\n",
    "  plt.legend()\n",
    "  plt.grid(True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>TotalPop</th>\n",
       "      <th>Men</th>\n",
       "      <th>Women</th>\n",
       "      <th>Hispanic</th>\n",
       "      <th>White</th>\n",
       "      <th>Black</th>\n",
       "      <th>Native</th>\n",
       "      <th>Asian</th>\n",
       "      <th>Pacific</th>\n",
       "      <th>Citizen</th>\n",
       "      <th>...</th>\n",
       "      <th>Transit</th>\n",
       "      <th>Walk</th>\n",
       "      <th>OtherTransp</th>\n",
       "      <th>WorkAtHome</th>\n",
       "      <th>MeanCommute</th>\n",
       "      <th>Employed</th>\n",
       "      <th>PrivateWork</th>\n",
       "      <th>PublicWork</th>\n",
       "      <th>SelfEmployed</th>\n",
       "      <th>FamilyWork</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1948</td>\n",
       "      <td>940</td>\n",
       "      <td>1008</td>\n",
       "      <td>0.9</td>\n",
       "      <td>87.4</td>\n",
       "      <td>7.7</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1503</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.5</td>\n",
       "      <td>2.3</td>\n",
       "      <td>2.1</td>\n",
       "      <td>25.0</td>\n",
       "      <td>943</td>\n",
       "      <td>77.1</td>\n",
       "      <td>18.3</td>\n",
       "      <td>4.6</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2156</td>\n",
       "      <td>1059</td>\n",
       "      <td>1097</td>\n",
       "      <td>0.8</td>\n",
       "      <td>40.4</td>\n",
       "      <td>53.3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1662</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>23.4</td>\n",
       "      <td>753</td>\n",
       "      <td>77.0</td>\n",
       "      <td>16.9</td>\n",
       "      <td>6.1</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2968</td>\n",
       "      <td>1364</td>\n",
       "      <td>1604</td>\n",
       "      <td>0.0</td>\n",
       "      <td>74.5</td>\n",
       "      <td>18.6</td>\n",
       "      <td>0.5</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.3</td>\n",
       "      <td>2335</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.5</td>\n",
       "      <td>19.6</td>\n",
       "      <td>1373</td>\n",
       "      <td>64.1</td>\n",
       "      <td>23.6</td>\n",
       "      <td>12.3</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4423</td>\n",
       "      <td>2172</td>\n",
       "      <td>2251</td>\n",
       "      <td>10.5</td>\n",
       "      <td>82.8</td>\n",
       "      <td>3.7</td>\n",
       "      <td>1.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3306</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.6</td>\n",
       "      <td>1.6</td>\n",
       "      <td>25.3</td>\n",
       "      <td>1782</td>\n",
       "      <td>75.7</td>\n",
       "      <td>21.2</td>\n",
       "      <td>3.1</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10763</td>\n",
       "      <td>4922</td>\n",
       "      <td>5841</td>\n",
       "      <td>0.7</td>\n",
       "      <td>68.5</td>\n",
       "      <td>24.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>7666</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.9</td>\n",
       "      <td>24.8</td>\n",
       "      <td>5037</td>\n",
       "      <td>67.1</td>\n",
       "      <td>27.6</td>\n",
       "      <td>5.3</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 28 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   TotalPop   Men  Women  Hispanic  White  Black  Native  Asian  Pacific   \n",
       "0      1948   940   1008       0.9   87.4    7.7     0.3    0.6      0.0  \\\n",
       "1      2156  1059   1097       0.8   40.4   53.3     0.0    2.3      0.0   \n",
       "2      2968  1364   1604       0.0   74.5   18.6     0.5    1.4      0.3   \n",
       "3      4423  2172   2251      10.5   82.8    3.7     1.6    0.0      0.0   \n",
       "4     10763  4922   5841       0.7   68.5   24.8     0.0    3.8      0.0   \n",
       "\n",
       "   Citizen  ...  Transit  Walk  OtherTransp  WorkAtHome  MeanCommute   \n",
       "0     1503  ...      0.0   0.5          2.3         2.1         25.0  \\\n",
       "1     1662  ...      0.0   0.0          0.7         0.0         23.4   \n",
       "2     2335  ...      0.0   0.0          0.0         2.5         19.6   \n",
       "3     3306  ...      0.0   0.0          2.6         1.6         25.3   \n",
       "4     7666  ...      0.0   0.0          0.6         0.9         24.8   \n",
       "\n",
       "   Employed  PrivateWork  PublicWork  SelfEmployed  FamilyWork  \n",
       "0       943         77.1        18.3           4.6         0.0  \n",
       "1       753         77.0        16.9           6.1         0.0  \n",
       "2      1373         64.1        23.6          12.3         0.0  \n",
       "3      1782         75.7        21.2           3.1         0.0  \n",
       "4      5037         67.1        27.6           5.3         0.0  \n",
       "\n",
       "[5 rows x 28 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = df.dropna()\n",
    "df = df[['TotalPop', 'Men', 'Women',\n",
    "       'Hispanic', 'White', 'Black', 'Native', 'Asian', 'Pacific', 'Citizen',\n",
    "       'IncomePerCap',\n",
    "       'Professional', 'Service', 'Office', 'Construction',\n",
    "       'Production', 'Drive', 'Carpool', 'Transit', 'Walk', 'OtherTransp',\n",
    "       'WorkAtHome', 'MeanCommute', 'Employed', 'PrivateWork', 'PublicWork',\n",
    "       'SelfEmployed', 'FamilyWork']]\n",
    "train_dataset = df.sample(frac=0.9, random_state=0)\n",
    "test_dataset = df.drop(train_dataset.index)\n",
    "\n",
    "train_features = train_dataset.copy()\n",
    "test_features = test_dataset.copy()\n",
    "\n",
    "train_labels = train_features.pop('IncomePerCap')\n",
    "test_labels = test_features.pop('IncomePerCap')\n",
    "\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "normalizer = tf.keras.layers.Normalization(axis=-1)\n",
    "normalizer.adapt(np.asarray(train_features).astype('float32'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "ann_model = tf.keras.Sequential([\n",
    "    normalizer,\n",
    "    tf.keras.layers.Dense(units=28, activation='relu'),\n",
    "    # tf.keras.layers.Dense(units=14, activation='relu'),\n",
    "    tf.keras.layers.Dense(units=7, activation='relu'),\n",
    "    tf.keras.layers.Dense(units=1, activation='linear')\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/150\n",
      "1179/1179 - 5s - loss: 81243680.0000 - val_loss: 48772696.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 2/150\n",
      "1179/1179 - 3s - loss: 48452296.0000 - val_loss: 52037948.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 3/150\n",
      "1179/1179 - 3s - loss: 46397564.0000 - val_loss: 44526424.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 4/150\n",
      "1179/1179 - 4s - loss: 44445776.0000 - val_loss: 43796928.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 5/150\n",
      "1179/1179 - 3s - loss: 43317748.0000 - val_loss: 52424120.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 6/150\n",
      "1179/1179 - 3s - loss: 42431768.0000 - val_loss: 41207544.0000 - 3s/epoch - 2ms/step\n",
      "Epoch 7/150\n",
      "1179/1179 - 4s - loss: 41758604.0000 - val_loss: 41322952.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 8/150\n",
      "1179/1179 - 3s - loss: 40596292.0000 - val_loss: 44070684.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 9/150\n",
      "1179/1179 - 4s - loss: 40599876.0000 - val_loss: 40598680.0000 - 4s/epoch - 4ms/step\n",
      "Epoch 10/150\n",
      "1179/1179 - 3s - loss: 40000348.0000 - val_loss: 43980572.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 11/150\n",
      "1179/1179 - 3s - loss: 39596428.0000 - val_loss: 38877664.0000 - 3s/epoch - 2ms/step\n",
      "Epoch 12/150\n",
      "1179/1179 - 3s - loss: 38903792.0000 - val_loss: 38014376.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 13/150\n",
      "1179/1179 - 3s - loss: 38474488.0000 - val_loss: 39760052.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 14/150\n",
      "1179/1179 - 4s - loss: 38504108.0000 - val_loss: 37364332.0000 - 4s/epoch - 4ms/step\n",
      "Epoch 15/150\n",
      "1179/1179 - 3s - loss: 37938512.0000 - val_loss: 37365228.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 16/150\n",
      "1179/1179 - 3s - loss: 37680316.0000 - val_loss: 37777060.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 17/150\n",
      "1179/1179 - 3s - loss: 37730892.0000 - val_loss: 39869068.0000 - 3s/epoch - 2ms/step\n",
      "Epoch 18/150\n",
      "1179/1179 - 4s - loss: 37245260.0000 - val_loss: 39766280.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 19/150\n",
      "1179/1179 - 4s - loss: 37046988.0000 - val_loss: 40039924.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 20/150\n",
      "1179/1179 - 3s - loss: 36707760.0000 - val_loss: 36432436.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 21/150\n",
      "1179/1179 - 3s - loss: 36681500.0000 - val_loss: 36219508.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 22/150\n",
      "1179/1179 - 3s - loss: 36432936.0000 - val_loss: 36140228.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 23/150\n",
      "1179/1179 - 4s - loss: 36373460.0000 - val_loss: 37884600.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 24/150\n",
      "1179/1179 - 3s - loss: 36186680.0000 - val_loss: 38323556.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 25/150\n",
      "1179/1179 - 4s - loss: 35900216.0000 - val_loss: 38717124.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 26/150\n",
      "1179/1179 - 3s - loss: 35674284.0000 - val_loss: 36445784.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 27/150\n",
      "1179/1179 - 3s - loss: 35844060.0000 - val_loss: 37424752.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 28/150\n",
      "1179/1179 - 3s - loss: 35858544.0000 - val_loss: 36248072.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 29/150\n",
      "1179/1179 - 3s - loss: 35523956.0000 - val_loss: 37450336.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 30/150\n",
      "1179/1179 - 3s - loss: 35342644.0000 - val_loss: 38168676.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 31/150\n",
      "1179/1179 - 3s - loss: 35303000.0000 - val_loss: 36697132.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 32/150\n",
      "1179/1179 - 3s - loss: 35200548.0000 - val_loss: 36204216.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 33/150\n",
      "1179/1179 - 3s - loss: 35779456.0000 - val_loss: 38131132.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 34/150\n",
      "1179/1179 - 4s - loss: 35514948.0000 - val_loss: 37891344.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 35/150\n",
      "1179/1179 - 3s - loss: 35005772.0000 - val_loss: 36105464.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 36/150\n",
      "1179/1179 - 3s - loss: 35253272.0000 - val_loss: 36141256.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 37/150\n",
      "1179/1179 - 3s - loss: 34731324.0000 - val_loss: 36645028.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 38/150\n",
      "1179/1179 - 3s - loss: 35372844.0000 - val_loss: 37371712.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 39/150\n",
      "1179/1179 - 3s - loss: 34633956.0000 - val_loss: 35695536.0000 - 3s/epoch - 2ms/step\n",
      "Epoch 40/150\n",
      "1179/1179 - 3s - loss: 35078436.0000 - val_loss: 36754560.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 41/150\n",
      "1179/1179 - 3s - loss: 34761056.0000 - val_loss: 36209812.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 42/150\n",
      "1179/1179 - 3s - loss: 34471536.0000 - val_loss: 38121228.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 43/150\n",
      "1179/1179 - 3s - loss: 34365284.0000 - val_loss: 37067252.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 44/150\n",
      "1179/1179 - 4s - loss: 34430224.0000 - val_loss: 37852320.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 45/150\n",
      "1179/1179 - 4s - loss: 34565616.0000 - val_loss: 39237328.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 46/150\n",
      "1179/1179 - 3s - loss: 34754404.0000 - val_loss: 38517952.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 47/150\n",
      "1179/1179 - 4s - loss: 34026908.0000 - val_loss: 37257580.0000 - 4s/epoch - 4ms/step\n",
      "Epoch 48/150\n",
      "1179/1179 - 3s - loss: 34196080.0000 - val_loss: 39255672.0000 - 3s/epoch - 2ms/step\n",
      "Epoch 49/150\n",
      "1179/1179 - 3s - loss: 33789628.0000 - val_loss: 40594252.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 50/150\n",
      "1179/1179 - 3s - loss: 34098612.0000 - val_loss: 37845612.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 51/150\n",
      "1179/1179 - 3s - loss: 34162032.0000 - val_loss: 36510924.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 52/150\n",
      "1179/1179 - 4s - loss: 33626508.0000 - val_loss: 37848520.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 53/150\n",
      "1179/1179 - 3s - loss: 34003704.0000 - val_loss: 37466872.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 54/150\n",
      "1179/1179 - 3s - loss: 34224008.0000 - val_loss: 36014932.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 55/150\n",
      "1179/1179 - 4s - loss: 33773932.0000 - val_loss: 41147312.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 56/150\n",
      "1179/1179 - 4s - loss: 33966584.0000 - val_loss: 36442332.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 57/150\n",
      "1179/1179 - 3s - loss: 33964252.0000 - val_loss: 36573664.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 58/150\n",
      "1179/1179 - 3s - loss: 34246944.0000 - val_loss: 36008392.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 59/150\n",
      "1179/1179 - 4s - loss: 33907676.0000 - val_loss: 39755572.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 60/150\n",
      "1179/1179 - 4s - loss: 33769876.0000 - val_loss: 38224452.0000 - 4s/epoch - 4ms/step\n",
      "Epoch 61/150\n",
      "1179/1179 - 4s - loss: 33781544.0000 - val_loss: 36913204.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 62/150\n",
      "1179/1179 - 4s - loss: 33706532.0000 - val_loss: 37199640.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 63/150\n",
      "1179/1179 - 4s - loss: 33664512.0000 - val_loss: 35217860.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 64/150\n",
      "1179/1179 - 4s - loss: 33009920.0000 - val_loss: 37319472.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 65/150\n",
      "1179/1179 - 4s - loss: 33457982.0000 - val_loss: 35952772.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 66/150\n",
      "1179/1179 - 3s - loss: 34112116.0000 - val_loss: 35697784.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 67/150\n",
      "1179/1179 - 3s - loss: 33459908.0000 - val_loss: 40180076.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 68/150\n",
      "1179/1179 - 4s - loss: 33543412.0000 - val_loss: 36044056.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 69/150\n",
      "1179/1179 - 6s - loss: 32929746.0000 - val_loss: 35797212.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 70/150\n",
      "1179/1179 - 6s - loss: 33332534.0000 - val_loss: 36081776.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 71/150\n",
      "1179/1179 - 6s - loss: 33509530.0000 - val_loss: 36177168.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 72/150\n",
      "1179/1179 - 6s - loss: 33163360.0000 - val_loss: 45957936.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 73/150\n",
      "1179/1179 - 6s - loss: 33453060.0000 - val_loss: 36648872.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 74/150\n",
      "1179/1179 - 6s - loss: 33389442.0000 - val_loss: 37614892.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 75/150\n",
      "1179/1179 - 6s - loss: 33641428.0000 - val_loss: 39990196.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 76/150\n",
      "1179/1179 - 5s - loss: 33385986.0000 - val_loss: 37687448.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 77/150\n",
      "1179/1179 - 6s - loss: 33462906.0000 - val_loss: 35593352.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 78/150\n",
      "1179/1179 - 5s - loss: 33201558.0000 - val_loss: 35811336.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 79/150\n",
      "1179/1179 - 6s - loss: 32966810.0000 - val_loss: 38729404.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 80/150\n",
      "1179/1179 - 3s - loss: 33906420.0000 - val_loss: 36476504.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 81/150\n",
      "1179/1179 - 6s - loss: 33461904.0000 - val_loss: 36885856.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 82/150\n",
      "1179/1179 - 5s - loss: 33922668.0000 - val_loss: 39566324.0000 - 5s/epoch - 5ms/step\n",
      "Epoch 83/150\n",
      "1179/1179 - 7s - loss: 32938152.0000 - val_loss: 42026892.0000 - 7s/epoch - 6ms/step\n",
      "Epoch 84/150\n",
      "1179/1179 - 5s - loss: 33130124.0000 - val_loss: 35213748.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 85/150\n",
      "1179/1179 - 4s - loss: 33008172.0000 - val_loss: 38513492.0000 - 4s/epoch - 4ms/step\n",
      "Epoch 86/150\n",
      "1179/1179 - 6s - loss: 33118266.0000 - val_loss: 35393976.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 87/150\n",
      "1179/1179 - 5s - loss: 33155484.0000 - val_loss: 34953924.0000 - 5s/epoch - 5ms/step\n",
      "Epoch 88/150\n",
      "1179/1179 - 5s - loss: 32915294.0000 - val_loss: 36566364.0000 - 5s/epoch - 5ms/step\n",
      "Epoch 89/150\n",
      "1179/1179 - 6s - loss: 32583538.0000 - val_loss: 37025956.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 90/150\n",
      "1179/1179 - 4s - loss: 33306380.0000 - val_loss: 35760124.0000 - 4s/epoch - 4ms/step\n",
      "Epoch 91/150\n",
      "1179/1179 - 5s - loss: 32755480.0000 - val_loss: 35094260.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 92/150\n",
      "1179/1179 - 6s - loss: 32695204.0000 - val_loss: 35296384.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 93/150\n",
      "1179/1179 - 5s - loss: 32825392.0000 - val_loss: 36585776.0000 - 5s/epoch - 5ms/step\n",
      "Epoch 94/150\n",
      "1179/1179 - 5s - loss: 32994088.0000 - val_loss: 36600696.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 95/150\n",
      "1179/1179 - 4s - loss: 32502518.0000 - val_loss: 35007992.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 96/150\n",
      "1179/1179 - 4s - loss: 32601434.0000 - val_loss: 38734448.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 97/150\n",
      "1179/1179 - 5s - loss: 32808868.0000 - val_loss: 36162140.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 98/150\n",
      "1179/1179 - 6s - loss: 32962390.0000 - val_loss: 35715696.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 99/150\n",
      "1179/1179 - 6s - loss: 32674752.0000 - val_loss: 37079428.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 100/150\n",
      "1179/1179 - 6s - loss: 33130420.0000 - val_loss: 36618960.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 101/150\n",
      "1179/1179 - 6s - loss: 32962078.0000 - val_loss: 35943756.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 102/150\n",
      "1179/1179 - 6s - loss: 32899468.0000 - val_loss: 35658576.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 103/150\n",
      "1179/1179 - 6s - loss: 32803864.0000 - val_loss: 35229196.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 104/150\n",
      "1179/1179 - 6s - loss: 32483298.0000 - val_loss: 37089984.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 105/150\n",
      "1179/1179 - 5s - loss: 32482854.0000 - val_loss: 34670436.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 106/150\n",
      "1179/1179 - 6s - loss: 32825502.0000 - val_loss: 35651168.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 107/150\n",
      "1179/1179 - 5s - loss: 32389638.0000 - val_loss: 35535164.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 108/150\n",
      "1179/1179 - 6s - loss: 32830040.0000 - val_loss: 37272620.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 109/150\n",
      "1179/1179 - 4s - loss: 32566598.0000 - val_loss: 35115216.0000 - 4s/epoch - 3ms/step\n",
      "Epoch 110/150\n",
      "1179/1179 - 3s - loss: 32365774.0000 - val_loss: 36994456.0000 - 3s/epoch - 3ms/step\n",
      "Epoch 111/150\n",
      "1179/1179 - 5s - loss: 32779034.0000 - val_loss: 34851788.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 112/150\n",
      "1179/1179 - 6s - loss: 32498576.0000 - val_loss: 36014660.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 113/150\n",
      "1179/1179 - 6s - loss: 32422388.0000 - val_loss: 37374556.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 114/150\n",
      "1179/1179 - 5s - loss: 32572196.0000 - val_loss: 35491596.0000 - 5s/epoch - 5ms/step\n",
      "Epoch 115/150\n",
      "1179/1179 - 6s - loss: 32451900.0000 - val_loss: 38359696.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 116/150\n",
      "1179/1179 - 5s - loss: 32376520.0000 - val_loss: 37709876.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 117/150\n",
      "1179/1179 - 5s - loss: 32437058.0000 - val_loss: 37681432.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 118/150\n",
      "1179/1179 - 5s - loss: 32625592.0000 - val_loss: 36782168.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 119/150\n",
      "1179/1179 - 6s - loss: 32440912.0000 - val_loss: 40070920.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 120/150\n",
      "1179/1179 - 6s - loss: 32503718.0000 - val_loss: 35201688.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 121/150\n",
      "1179/1179 - 6s - loss: 32485094.0000 - val_loss: 36027184.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 122/150\n",
      "1179/1179 - 5s - loss: 32217106.0000 - val_loss: 36274468.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 123/150\n",
      "1179/1179 - 6s - loss: 32766732.0000 - val_loss: 40126460.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 124/150\n",
      "1179/1179 - 5s - loss: 32497932.0000 - val_loss: 36787472.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 125/150\n",
      "1179/1179 - 5s - loss: 32419950.0000 - val_loss: 36134808.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 126/150\n",
      "1179/1179 - 5s - loss: 32348582.0000 - val_loss: 37830768.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 127/150\n",
      "1179/1179 - 5s - loss: 31978240.0000 - val_loss: 36794216.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 128/150\n",
      "1179/1179 - 6s - loss: 32380960.0000 - val_loss: 37144840.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 129/150\n",
      "1179/1179 - 6s - loss: 32221586.0000 - val_loss: 36286796.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 130/150\n",
      "1179/1179 - 6s - loss: 32122140.0000 - val_loss: 37914148.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 131/150\n",
      "1179/1179 - 6s - loss: 32391438.0000 - val_loss: 36137632.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 132/150\n",
      "1179/1179 - 6s - loss: 32089252.0000 - val_loss: 35807776.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 133/150\n",
      "1179/1179 - 6s - loss: 32320812.0000 - val_loss: 35367784.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 134/150\n",
      "1179/1179 - 6s - loss: 32312696.0000 - val_loss: 35953484.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 135/150\n",
      "1179/1179 - 6s - loss: 32706264.0000 - val_loss: 36072036.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 136/150\n",
      "1179/1179 - 6s - loss: 31957462.0000 - val_loss: 35887516.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 137/150\n",
      "1179/1179 - 6s - loss: 32043280.0000 - val_loss: 36626124.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 138/150\n",
      "1179/1179 - 6s - loss: 32291004.0000 - val_loss: 35153736.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 139/150\n",
      "1179/1179 - 6s - loss: 32406242.0000 - val_loss: 36310700.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 140/150\n",
      "1179/1179 - 6s - loss: 32242770.0000 - val_loss: 35796908.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 141/150\n",
      "1179/1179 - 6s - loss: 32063750.0000 - val_loss: 35372924.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 142/150\n",
      "1179/1179 - 5s - loss: 32050986.0000 - val_loss: 34838652.0000 - 5s/epoch - 5ms/step\n",
      "Epoch 143/150\n",
      "1179/1179 - 6s - loss: 32286098.0000 - val_loss: 35115268.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 144/150\n",
      "1179/1179 - 6s - loss: 31989062.0000 - val_loss: 36328016.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 145/150\n",
      "1179/1179 - 5s - loss: 32245144.0000 - val_loss: 36198916.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 146/150\n",
      "1179/1179 - 6s - loss: 32380650.0000 - val_loss: 38630340.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 147/150\n",
      "1179/1179 - 6s - loss: 32078010.0000 - val_loss: 35502488.0000 - 6s/epoch - 5ms/step\n",
      "Epoch 148/150\n",
      "1179/1179 - 5s - loss: 32228030.0000 - val_loss: 34926544.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 149/150\n",
      "1179/1179 - 5s - loss: 32268590.0000 - val_loss: 35323516.0000 - 5s/epoch - 4ms/step\n",
      "Epoch 150/150\n",
      "1179/1179 - 4s - loss: 32023880.0000 - val_loss: 39020620.0000 - 4s/epoch - 4ms/step\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ann_model.compile(\n",
    "    optimizer=tf.keras.optimizers.Adam(learning_rate=0.1),\n",
    "    loss='mean_squared_error'\n",
    "    )\n",
    "\n",
    "history = ann_model.fit(\n",
    "    train_features,\n",
    "    train_labels,\n",
    "    epochs=150,\n",
    "    batch_size=50,\n",
    "    validation_split=0.1,\n",
    "    verbose=2\n",
    ")\n",
    "plot_loss(history)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "RESULTS:\n",
    "\n",
    "-----------------------------------------\n",
    "\n",
    "Using just 2015 data\n",
    "\n",
    "1 layer  : 34435316.0000 MSE\n",
    "\n",
    "2 layers : 32542424.0000 MSE (28, 14, 1)\n",
    "\n",
    "2 layers : 31465028.0000 MSE (28,  7, 1)\n",
    "\n",
    "3 layers : 33440750.0000 MSE\n",
    "\n",
    "------------------------------------------\n",
    "\n",
    "Using 2015 + 2017 data\n",
    "\n",
    "1 layer  : 49900776.0000 MSE\n",
    "\n",
    "2 layers : 36901260.0000 MSE (28, 14, 1)\n",
    "\n",
    "2 layers : 37744916.0000 MSE (28,  7, 1)\n",
    "\n",
    "3 layers : 40632372.0000 MSE\n",
    "\n",
    "-----------------------------------------"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now how does the best model trained on 2015 data perform on the 2017 data?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "newdf = pd.read_csv('archive/acs2017_census_tract_data.csv').rename(columns={'VotingAgeCitizen':'Citizen'})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def mse(y, y_prime, ignore_size_mismatch=False):\n",
    "    if (len(y) != len(y_prime)) and not ignore_size_mismatch:\n",
    "        print(len(y), len(y_prime))\n",
    "        raise ValueError(\"Mismatched lengths\")\n",
    "    sum = 0\n",
    "    for i in range(min(len(y), len(y_prime))):\n",
    "        sum += (y[i] - y_prime[i])*(y[i] - y_prime[i])\n",
    "    return (sum/min(len(y), len(y_prime)))[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2283/2283 [==============================] - 8s 3ms/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "378935453.6764867"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_prime = ann_model.predict(newdf[['TotalPop', 'Men', 'Women',\n",
    "       'Hispanic', 'White', 'Black', 'Native', 'Asian', 'Pacific', 'Citizen',\n",
    "       'Professional', 'Service', 'Office', 'Construction',\n",
    "       'Production', 'Drive', 'Carpool', 'Transit', 'Walk', 'OtherTransp',\n",
    "       'WorkAtHome', 'MeanCommute', 'Employed', 'PrivateWork', 'PublicWork',\n",
    "       'SelfEmployed', 'FamilyWork']].dropna())\n",
    "y = newdf['IncomePerCap'].dropna().to_numpy()\n",
    "# print(\"Y:\",y,\"\\nY':\", y_prime)\n",
    "# print(y.shape, y_prime.shape)\n",
    "mse(y, y_prime, ignore_size_mismatch=True)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Decent! But still significantly worse than on the original dataset."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So where is the error the highest when using the old data to predict the new? Why?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from queue import PriorityQueue\n",
    "\n",
    "def get_n_highest_error(df, model, n):\n",
    "   y_prime = model.predict(df[['TotalPop', 'Men', 'Women',\n",
    "       'Hispanic', 'White', 'Black', 'Native', 'Asian', 'Pacific', 'Citizen',\n",
    "       'Professional', 'Service', 'Office', 'Construction',\n",
    "       'Production', 'Drive', 'Carpool', 'Transit', 'Walk', 'OtherTransp',\n",
    "       'WorkAtHome', 'MeanCommute', 'Employed', 'PrivateWork', 'PublicWork',\n",
    "       'SelfEmployed', 'FamilyWork']].dropna())\n",
    "   y = df['IncomePerCap'].dropna().to_numpy()\n",
    "   error_q = PriorityQueue()\n",
    "   for i in range(min(len(y), len(y_prime))):\n",
    "      error = (y[i] - y_prime[i])*(y[i] - y_prime[i])\n",
    "      error_q.put((error, i))\n",
    "   max_error_rows = []\n",
    "   for _ in range(n):\n",
    "      max_error_rows.append(df.iloc[error_q.get()[1]])\n",
    "   return pd.DataFrame(max_error_rows)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2283/2283 [==============================] - 7s 3ms/step\n"
     ]
    }
   ],
   "source": [
    "max_error_df = get_n_highest_error(df=newdf, model=ann_model, n=len(newdf)//4) #look at the top quarter most poorly predicted values\n",
    "# print(max_error_df.head(10))\n",
    "max_error_df_desc = max_error_df.describe()\n",
    "newdf_desc = newdf.describe()\n",
    "df_desc = df.describe()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "But first, how does the 2017 data differ from the 2015 data?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Percent change in value from 2015 to 2017\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Asian             -3.574975\n",
       "Black             -0.374849\n",
       "Carpool            3.414935\n",
       "ChildPoverty            NaN\n",
       "Citizen           -0.728458\n",
       "Construction       0.785187\n",
       "Drive              0.017708\n",
       "Employed          -1.677136\n",
       "FamilyWork        -0.622012\n",
       "Hispanic          -2.330014\n",
       "Income                  NaN\n",
       "IncomeErr               NaN\n",
       "IncomePerCap      -7.471870\n",
       "IncomePerCapErr         NaN\n",
       "MeanCommute       -1.426809\n",
       "Men               -0.178896\n",
       "Native            -1.533357\n",
       "Office             2.083557\n",
       "OtherTransp       -0.756946\n",
       "Pacific           -2.487061\n",
       "Poverty                 NaN\n",
       "PrivateWork       -0.617113\n",
       "Production        -0.112546\n",
       "Professional      -2.183607\n",
       "PublicWork         2.990662\n",
       "SelfEmployed       0.838041\n",
       "Service            1.061951\n",
       "TotalPop          -0.026505\n",
       "TractId                 NaN\n",
       "Transit            0.836002\n",
       "Unemployment            NaN\n",
       "Walk              -1.152718\n",
       "White              1.216423\n",
       "Women              0.120703\n",
       "WorkAtHome        -7.845344\n",
       "Name: mean, dtype: float64"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_minus_newdf = 100*(df_desc - newdf_desc)/(df_desc)\n",
    "print(\"Percent change in value from 2015 to 2017\")\n",
    "df_minus_newdf.iloc[1]"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Answer: They don't seem to differ too much - the largest differences are income per capita and the percentage of people working at home.\n",
    "\n",
    "Now, let's see how the rows our 2015 model struggled with differ from the rest of the 2017 dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "TractId            -5.126066\n",
       "TotalPop           -0.095182\n",
       "Men                -0.073815\n",
       "Women              -0.115885\n",
       "Hispanic           -8.646638\n",
       "White               1.836139\n",
       "Black               4.520187\n",
       "Native             15.701823\n",
       "Asian              -7.508631\n",
       "Pacific             3.629341\n",
       "Citizen            -0.066504\n",
       "Income             -1.933477\n",
       "IncomeErr          -2.007968\n",
       "IncomePerCap       -2.293444\n",
       "IncomePerCapErr    -2.454678\n",
       "Poverty             0.187865\n",
       "ChildPoverty        0.949416\n",
       "Professional       -2.090230\n",
       "Service            -0.348051\n",
       "Office             -0.231473\n",
       "Construction        2.306312\n",
       "Production          5.034000\n",
       "Drive               1.032715\n",
       "Carpool             1.563397\n",
       "Transit           -11.525867\n",
       "Walk               -5.468805\n",
       "OtherTransp        -2.333750\n",
       "WorkAtHome         -2.036524\n",
       "MeanCommute        -1.011118\n",
       "Employed           -0.402619\n",
       "PrivateWork         0.223898\n",
       "PublicWork         -0.736945\n",
       "SelfEmployed       -1.269197\n",
       "FamilyWork          2.951608\n",
       "Unemployment       -0.434305\n",
       "Name: mean, dtype: float64"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "error_minus_newdf = 100*(max_error_df_desc - newdf_desc)/(newdf_desc)\n",
    "error_minus_newdf.iloc[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              PercentageInHighErrorData\n",
      "State                                  \n",
      "Alabama                        0.729890\n",
      "Wyoming                        0.484848\n",
      "Arizona                        0.395806\n",
      "South Dakota                   0.387387\n",
      "Mississippi                    0.384036\n",
      "Arkansas                       0.374636\n",
      "Wisconsin                      0.343506\n",
      "Hawaii                         0.324786\n",
      "Rhode Island                   0.319672\n",
      "Indiana                        0.307743\n"
     ]
    }
   ],
   "source": [
    "state_df = pd.DataFrame(max_error_df.groupby('State').count()['TractId']/newdf.groupby('State').count()['TractId']).rename(columns={'TractId':'PercentageInHighErrorData'})\n",
    "print(state_df.sort_values(by='PercentageInHighErrorData',ascending=False).head(10))"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Another way to see this..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.plotly.v1+json": {
       "config": {
        "plotlyServerURL": "https://plot.ly"
       },
       "data": [
        {
         "coloraxis": "coloraxis",
         "geo": "geo",
         "hovertemplate": "Abbreviation=%{location}<br>PercentageInHighErrorData=%{z}<extra></extra>",
         "locationmode": "USA-states",
         "locations": [
          "AL",
          "AK",
          "AZ",
          "AR",
          "CA",
          "CO",
          "CT",
          "DE",
          "DC",
          "FL",
          "GA",
          "HI",
          "ID",
          "IL",
          "IN",
          "IA",
          "KS",
          "KY",
          "LA",
          "ME",
          "MD",
          "MA",
          "MI",
          "MN",
          "MS",
          "MO",
          "MT",
          "NE",
          "NV",
          "NH",
          "NJ",
          "NM",
          "NY",
          "NC",
          "ND",
          "OH",
          "OK",
          "OR",
          "PA",
          "RI",
          "SC",
          "SD",
          "TN",
          "TX",
          "UT",
          "VT",
          "VA",
          "WA",
          "WV",
          "WI",
          "WY"
         ],
         "name": "",
         "type": "choropleth",
         "z": [
          0.7298899237933955,
          0.16766467065868262,
          0.39580602883355176,
          0.3746355685131195,
          0.23296512349509743,
          0.17133706965572457,
          0.15006002400960383,
          0.13302752293577982,
          0.24581005586592178,
          0.224263839811543,
          0.2493651599796851,
          0.3247863247863248,
          0.2516778523489933,
          0.2308677553634326,
          0.30774321641297153,
          0.3054545454545455,
          0.2922077922077922,
          0.29596412556053814,
          0.2883275261324042,
          0.19832402234636873,
          0.21408250355618777,
          0.20974289580514208,
          0.27479559189477426,
          0.23692077727952168,
          0.38403614457831325,
          0.26346015793251976,
          0.22509225092250923,
          0.2387218045112782,
          0.1804949053857351,
          0.15254237288135594,
          0.16567164179104477,
          0.2625250501002004,
          0.21004473363155754,
          0.2223234624145786,
          0.2146341463414634,
          0.2638888888888889,
          0.29732313575525815,
          0.2278177458033573,
          0.24549409571162212,
          0.319672131147541,
          0.3073436083408885,
          0.38738738738738737,
          0.2531730126920508,
          0.22564102564102564,
          0.1445578231292517,
          0.08152173913043478,
          0.2108023072889355,
          0.2956104252400549,
          0.2128099173553719,
          0.3435060326472676,
          0.48484848484848486
         ]
        }
       ],
       "layout": {
        "coloraxis": {
         "colorbar": {
          "title": {
           "text": "PercentageInHighErrorData"
          }
         },
         "colorscale": [
          [
           0,
           "#fde725"
          ],
          [
           0.1111111111111111,
           "#b5de2b"
          ],
          [
           0.2222222222222222,
           "#6ece58"
          ],
          [
           0.3333333333333333,
           "#35b779"
          ],
          [
           0.4444444444444444,
           "#1f9e89"
          ],
          [
           0.5555555555555556,
           "#26828e"
          ],
          [
           0.6666666666666666,
           "#31688e"
          ],
          [
           0.7777777777777778,
           "#3e4989"
          ],
          [
           0.8888888888888888,
           "#482878"
          ],
          [
           1,
           "#440154"
          ]
         ]
        },
        "geo": {
         "center": {},
         "domain": {
          "x": [
           0,
           1
          ],
          "y": [
           0,
           1
          ]
         },
         "scope": "usa"
        },
        "legend": {
         "tracegroupgap": 0
        },
        "margin": {
         "t": 60
        },
        "template": {
         "data": {
          "bar": [
           {
            "error_x": {
             "color": "#2a3f5f"
            },
            "error_y": {
             "color": "#2a3f5f"
            },
            "marker": {
             "line": {
              "color": "#E5ECF6",
              "width": 0.5
             },
             "pattern": {
              "fillmode": "overlay",
              "size": 10,
              "solidity": 0.2
             }
            },
            "type": "bar"
           }
          ],
          "barpolar": [
           {
            "marker": {
             "line": {
              "color": "#E5ECF6",
              "width": 0.5
             },
             "pattern": {
              "fillmode": "overlay",
              "size": 10,
              "solidity": 0.2
             }
            },
            "type": "barpolar"
           }
          ],
          "carpet": [
           {
            "aaxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "white",
             "linecolor": "white",
             "minorgridcolor": "white",
             "startlinecolor": "#2a3f5f"
            },
            "baxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "white",
             "linecolor": "white",
             "minorgridcolor": "white",
             "startlinecolor": "#2a3f5f"
            },
            "type": "carpet"
           }
          ],
          "choropleth": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "choropleth"
           }
          ],
          "contour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "contour"
           }
          ],
          "contourcarpet": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "contourcarpet"
           }
          ],
          "heatmap": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmap"
           }
          ],
          "heatmapgl": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmapgl"
           }
          ],
          "histogram": [
           {
            "marker": {
             "pattern": {
              "fillmode": "overlay",
              "size": 10,
              "solidity": 0.2
             }
            },
            "type": "histogram"
           }
          ],
          "histogram2d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2d"
           }
          ],
          "histogram2dcontour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2dcontour"
           }
          ],
          "mesh3d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "mesh3d"
           }
          ],
          "parcoords": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "parcoords"
           }
          ],
          "pie": [
           {
            "automargin": true,
            "type": "pie"
           }
          ],
          "scatter": [
           {
            "fillpattern": {
             "fillmode": "overlay",
             "size": 10,
             "solidity": 0.2
            },
            "type": "scatter"
           }
          ],
          "scatter3d": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatter3d"
           }
          ],
          "scattercarpet": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattercarpet"
           }
          ],
          "scattergeo": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergeo"
           }
          ],
          "scattergl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergl"
           }
          ],
          "scattermapbox": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattermapbox"
           }
          ],
          "scatterpolar": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolar"
           }
          ],
          "scatterpolargl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolargl"
           }
          ],
          "scatterternary": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterternary"
           }
          ],
          "surface": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "surface"
           }
          ],
          "table": [
           {
            "cells": {
             "fill": {
              "color": "#EBF0F8"
             },
             "line": {
              "color": "white"
             }
            },
            "header": {
             "fill": {
              "color": "#C8D4E3"
             },
             "line": {
              "color": "white"
             }
            },
            "type": "table"
           }
          ]
         },
         "layout": {
          "annotationdefaults": {
           "arrowcolor": "#2a3f5f",
           "arrowhead": 0,
           "arrowwidth": 1
          },
          "autotypenumbers": "strict",
          "coloraxis": {
           "colorbar": {
            "outlinewidth": 0,
            "ticks": ""
           }
          },
          "colorscale": {
           "diverging": [
            [
             0,
             "#8e0152"
            ],
            [
             0.1,
             "#c51b7d"
            ],
            [
             0.2,
             "#de77ae"
            ],
            [
             0.3,
             "#f1b6da"
            ],
            [
             0.4,
             "#fde0ef"
            ],
            [
             0.5,
             "#f7f7f7"
            ],
            [
             0.6,
             "#e6f5d0"
            ],
            [
             0.7,
             "#b8e186"
            ],
            [
             0.8,
             "#7fbc41"
            ],
            [
             0.9,
             "#4d9221"
            ],
            [
             1,
             "#276419"
            ]
           ],
           "sequential": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ],
           "sequentialminus": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ]
          },
          "colorway": [
           "#636efa",
           "#EF553B",
           "#00cc96",
           "#ab63fa",
           "#FFA15A",
           "#19d3f3",
           "#FF6692",
           "#B6E880",
           "#FF97FF",
           "#FECB52"
          ],
          "font": {
           "color": "#2a3f5f"
          },
          "geo": {
           "bgcolor": "white",
           "lakecolor": "white",
           "landcolor": "#E5ECF6",
           "showlakes": true,
           "showland": true,
           "subunitcolor": "white"
          },
          "hoverlabel": {
           "align": "left"
          },
          "hovermode": "closest",
          "mapbox": {
           "style": "light"
          },
          "paper_bgcolor": "white",
          "plot_bgcolor": "#E5ECF6",
          "polar": {
           "angularaxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           },
           "bgcolor": "#E5ECF6",
           "radialaxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           }
          },
          "scene": {
           "xaxis": {
            "backgroundcolor": "#E5ECF6",
            "gridcolor": "white",
            "gridwidth": 2,
            "linecolor": "white",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "white"
           },
           "yaxis": {
            "backgroundcolor": "#E5ECF6",
            "gridcolor": "white",
            "gridwidth": 2,
            "linecolor": "white",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "white"
           },
           "zaxis": {
            "backgroundcolor": "#E5ECF6",
            "gridcolor": "white",
            "gridwidth": 2,
            "linecolor": "white",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "white"
           }
          },
          "shapedefaults": {
           "line": {
            "color": "#2a3f5f"
           }
          },
          "ternary": {
           "aaxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           },
           "baxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           },
           "bgcolor": "#E5ECF6",
           "caxis": {
            "gridcolor": "white",
            "linecolor": "white",
            "ticks": ""
           }
          },
          "title": {
           "x": 0.05
          },
          "xaxis": {
           "automargin": true,
           "gridcolor": "white",
           "linecolor": "white",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "white",
           "zerolinewidth": 2
          },
          "yaxis": {
           "automargin": true,
           "gridcolor": "white",
           "linecolor": "white",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "white",
           "zerolinewidth": 2
          }
         }
        }
       }
      }
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import plotly.express as px\n",
    "statenames_to_codes = pd.read_csv('archive/statenamestocodes.csv')\n",
    "tmp = pd.merge(state_df, statenames_to_codes, on=\"State\")\n",
    "fig = px.choropleth(tmp,\n",
    "                    locations='Abbreviation', \n",
    "                    locationmode=\"USA-states\", \n",
    "                    color='PercentageInHighErrorData',\n",
    "                    color_continuous_scale=\"Viridis_r\", \n",
    "                    scope=\"usa\")\n",
    "fig.show()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "What on earth is up with Alabama?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Asian               2.226540\n",
      "Black               1.134586\n",
      "Carpool            -6.568011\n",
      "CensusTract              NaN\n",
      "ChildPoverty       -4.310358\n",
      "Citizen             0.853909\n",
      "Construction       -1.025833\n",
      "Drive               0.611688\n",
      "Employed            1.640884\n",
      "FamilyWork          1.654646\n",
      "Hispanic            2.540855\n",
      "Income              6.101793\n",
      "IncomeErr           7.908944\n",
      "IncomePerCap        6.847178\n",
      "IncomePerCapErr    10.717739\n",
      "MeanCommute         1.013335\n",
      "Men                 0.414892\n",
      "Native             -0.262664\n",
      "Office             -1.555397\n",
      "OtherTransp         0.812540\n",
      "Pacific           -16.336634\n",
      "Poverty            -4.123328\n",
      "PrivateWork         1.132186\n",
      "Production          0.808438\n",
      "Professional        1.618858\n",
      "PublicWork         -4.967478\n",
      "SelfEmployed       -1.577517\n",
      "Service            -0.863786\n",
      "TotalPop            0.417151\n",
      "TractId                  NaN\n",
      "Transit            -5.599436\n",
      "Unemployment      -19.043713\n",
      "Walk               -0.930205\n",
      "White              -0.931712\n",
      "Women               0.419276\n",
      "WorkAtHome          6.166271\n",
      "Name: mean, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "df = pd.read_csv('archive/acs2015_census_tract_data.csv')\n",
    "newdf = pd.read_csv('archive/acs2017_census_tract_data.csv').rename(columns={'VotingAgeCitizen':'Citizen'})\n",
    "\n",
    "alabama_2015 = df[df['State']=='Alabama']\n",
    "alabama_2017 = newdf[newdf['State']=='Alabama']\n",
    "\n",
    "print((100*(alabama_2017.describe()-alabama_2015.describe())/alabama_2015.describe()).iloc[1,:])"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Hard to say from this data: It might just be a matter of shifting income and poverty with otherwise consistent demographics.\n",
    "\n",
    "Let's try the second-most problematic state, Wyoming."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Asian              -2.815700\n",
      "Black              -5.604076\n",
      "Carpool            -8.387405\n",
      "CensusTract              NaN\n",
      "ChildPoverty      -15.231241\n",
      "Citizen             0.648333\n",
      "Construction       -4.343320\n",
      "Drive               1.429288\n",
      "Employed           -0.107502\n",
      "FamilyWork          6.896552\n",
      "Hispanic            0.102580\n",
      "Income              3.403878\n",
      "IncomeErr           7.596455\n",
      "IncomePerCap        4.949291\n",
      "IncomePerCapErr    10.537099\n",
      "MeanCommute        -1.801132\n",
      "Men                 0.927051\n",
      "Native              6.562500\n",
      "Office             -3.024348\n",
      "OtherTransp        -8.749122\n",
      "Pacific            -2.127660\n",
      "Poverty            -4.730023\n",
      "PrivateWork         0.325680\n",
      "Production          1.646065\n",
      "Professional        2.757133\n",
      "PublicWork         -1.237263\n",
      "SelfEmployed        0.263429\n",
      "Service             1.334764\n",
      "TotalPop            0.607405\n",
      "TractId                  NaN\n",
      "Transit           -15.675116\n",
      "Unemployment       -5.016229\n",
      "Walk                2.665951\n",
      "White               0.017214\n",
      "Women               0.274886\n",
      "WorkAtHome          3.320095\n",
      "Name: mean, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "wyoming_2015 = df[df['State']=='Wyoming']\n",
    "wyoming_2017 = newdf[newdf['State']=='Wyoming']\n",
    "\n",
    "print((100*(wyoming_2017.describe()-wyoming_2015.describe())/wyoming_2015.describe()).iloc[1,:])"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There doesn't appear to be much similarity here, but Wyoming is a very underpopulated state so it would make sense that the model might struggle because it might be more quirky in its demographics and easily swayed."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In sum, a neural net performs very well with this dataset. Simpler seems to be better - both because of data and because of the relationships between features and the label. \n",
    "\n",
    "There is something odd however about Alabama across the time between 2015 and 2017, and it would be interesting to continue exploring that."
   ]
  }
 ],
 "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": "3.11.2"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}