Bridges.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "\"\"\"Imports necessary packages\"\"\"\n",
    "%matplotlib inline\n",
    "import matplotlib\n",
    "import seaborn as sns\n",
    "matplotlib.rcParams['savefig.dpi'] = 2 * matplotlib.rcParams['savefig.dpi']\n",
    "from matplotlib import cm\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>ITEM NO</th>\n",
       "      <th>ITEM NAME</th>\n",
       "      <th>ITEM POSITION</th>\n",
       "      <th>ITEM LENGTH/TYPE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>State Code</td>\n",
       "      <td>1 - 3</td>\n",
       "      <td>3/N</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>Structure Number</td>\n",
       "      <td>4 - 18</td>\n",
       "      <td>15/AN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>Inventory Route</td>\n",
       "      <td>19 - 27</td>\n",
       "      <td>9/AN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>5A</td>\n",
       "      <td>Record Type</td>\n",
       "      <td>19</td>\n",
       "      <td>1/AN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>5B</td>\n",
       "      <td>Route Signing Prefix</td>\n",
       "      <td>20</td>\n",
       "      <td>1/N</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0 ITEM NO             ITEM NAME ITEM POSITION ITEM LENGTH/TYPE\n",
       "0           0       1            State Code         1 - 3              3/N\n",
       "1           1       8      Structure Number        4 - 18            15/AN\n",
       "2           2       5       Inventory Route       19 - 27             9/AN\n",
       "3           3      5A           Record Type           19              1/AN\n",
       "4           4      5B  Route Signing Prefix            20              1/N"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "schema = pd.read_csv(\"resources/nbi_record_format.csv\")\n",
    "schema.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 State Code\n",
      "8 Structure Number\n",
      "5 Inventory Route\n",
      "5A Record Type\n",
      "5B Route Signing Prefix\n",
      "5C Designated Level of Service\n",
      "5D Route Number\n",
      "5E Directional Suffix\n",
      "2 Highway Agency District\n",
      "3 County (Parish) Code\n",
      "4 Place Code\n",
      "6 Features Intersected\n",
      "6A Features Intersected\n",
      "6B Critical Facility Indicator\n",
      "7 Facility Carried By Structure\n",
      "9 Location\n",
      "10 Inventory Rte, Min Vert Clearance\n",
      "11 Kilometerpoint\n",
      "12 Base Highway Network\n",
      "13 Inventory Route, Subroute Number\n",
      "13A LRS Inventory Route\n",
      "13B Subroute Number\n",
      "16 Latitude\n",
      "17 Longitude\n",
      "19 Bypass/Detour Length\n",
      "20 Toll\n",
      "21 Maintenance Responsibility\n",
      "22 Owner\n",
      "26 Functional Class Of Inventory Rte.\n",
      "27 Year Built\n",
      "28 Lanes On/Under Structure\n",
      "28A Lanes On Structure\n",
      "28B Lanes Under Structure\n",
      "29 Average Daily Traffic\n",
      "30 Year Of Average Daily Traffic\n",
      "31 Design Load\n",
      "32 Approach Roadway Width\n",
      "33 Bridge Median\n",
      "34 Skew\n",
      "35 Structure Flared\n",
      "36 Traffic Safety Features\n",
      "36A Bridge Railings\n",
      "36B Transitions\n",
      "36C Approach Guardrail\n",
      "36D Approach Guardrail Ends\n",
      "37 Historical significance\n",
      "38 Navigation Control\n",
      "39 Navigation Vertical Clearance\n",
      "40 Navigation Horizontal Clearance\n",
      "41 Structure Open/Posted/Closed\n",
      "42 Type Of Service\n",
      "42A Type of Service On Bridge\n",
      "42B Type of Service Under Bridge\n",
      "43 Structure Type, Main\n",
      "43A Kind of Material/Design\n",
      "43B Type of Design/Construction\n",
      "44 Structure Type, Approach Spans\n",
      "44A Kind of Material/Design\n",
      "44B Type of Design/Construction\n",
      "45 Number Of Spans In Main Unit\n",
      "46 Number Of Approach Spans\n",
      "47 Inventory Rte Total Horz Clearance\n",
      "48 Length Of Maximum Span\n",
      "49 Structure Length\n",
      "50 Curb/Sidewalk Widths\n",
      "50A Left Curb/Sidewalk Width\n",
      "50B Right Curb/Sidewalk Width\n",
      "51 Bridge Roadway Width Curb-To-Curb\n",
      "52 Deck Width, Out-To-Out\n",
      "53 Min Vert Clear Over Bridge Roadway\n",
      "54 Minimum Vertical Underclearance\n",
      "54A Reference Feature\n",
      "54B Minimum Vertical Underclearance\n",
      "55 Min Lateral Underclear On Right\n",
      "55A Reference Feature\n",
      "55B Minimum Lateral Underclearance\n",
      "56 Min Lateral Underclear On Left\n",
      "58 Deck\n",
      "59 Superstructure\n",
      "60 Substructure\n",
      "61 Channel/Channel Protection\n",
      "62 Culverts\n",
      "63 Method Used To Determine Operating Rating\n",
      "64 Operating Rating\n",
      "65 Method Used To Determine Inventory Rating\n",
      "66 Inventory Rating\n",
      "67 Structural Evaluation\n",
      "68 Deck Geometry\n",
      "69 Underclear, Vertical & Horizontal\n",
      "70 Bridge Posting\n",
      "71 Waterway Adequacy\n",
      "72 Approach Roadway Alignment\n",
      "75 Type of Work\n",
      "75A Type of Work Proposed\n",
      "75B Work Done By\n",
      "76 Length Of Structure Improvement\n",
      "90 Inspection Date\n",
      "91 Designated Inspection Frequency\n",
      "92 Critical Feature Inspection\n",
      "92A Fracture Critical Details\n",
      "92B Underwater Inspection\n",
      "92C Other Special Inspection\n",
      "93 Critical Feature Inspection Dates\n",
      "93A Fracture Critical Details Date\n",
      "93B Underwater Inspection Date\n",
      "93C Other Special Inspection Date\n",
      "94 Bridge Improvement Cost\n",
      "95 Roadway Improvement Cost\n",
      "96 Total Project Cost\n",
      "97 Year Of Improvement Cost Estimate\n",
      "98 Border Bridge\n",
      "98A Neighboring State Code\n",
      "98B Percent Responsibility\n",
      "99 Border Bridge Structure Number\n",
      "100 STRAHNET Highway Designation\n",
      "101 Parallel Structure Designation\n",
      "102 Direction Of Traffic\n",
      "103 Temporary Structure Designation\n",
      "104 Highway System Of Inventory Route\n",
      "105 Federal Lands Highways\n",
      "106 Year Reconstructed\n",
      "107 Deck Structure Type\n",
      "108 Wearing Surface/Protective System\n",
      "108A Type of Wearing Surface\n",
      "108B Type of Membrane\n",
      "108C Deck Protection\n",
      "109 AVERAGE DAILY TRUCK TRAFFIC\n",
      "110 DESIGNATED NATIONAL NETWORK\n",
      "111 PIER/ABUTMENT PROTECTION\n",
      "112 NBIS BRIDGE LENGTH\n",
      "113 SCOUR CRITICAL BRIDGES\n",
      "114 FUTURE AVERAGE DAILY TRAFFIC\n",
      "115 YEAR OF FUTURE AVG DAILY TRAFFIC\n",
      "116 MINIMUM NAVIGATION VERTICAL CLEARANCE VERTICAL LIFT BRIDGE\n",
      "  FEDERAL AGENCY INDICATOR\n",
      "  Washington Headquarters Use\n",
      "  STATUS with 10yr rule\n",
      "n/a Asterisk Field in SR\n",
      "SR SUFFICIENCY RATING (select from last 4 positions only)\n",
      "nan STATUS without 10yr rule. Field new in 2013 and future years (This field is not part of the NBI annual file submittal. It is a download field only.)\n"
     ]
    }
   ],
   "source": [
    "for row in schema[[\"ITEM NO\", \"ITEM NAME\"]].itertuples():\n",
    "  print(row._1, row._2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/joshuaarnold/anaconda/lib/python3.5/site-packages/IPython/core/interactiveshell.py:2717: DtypeWarning: Columns (1,2,5,7,17,31,36,37,38,39,41,44,61,63,66,67,68,69,70,71,73,75,76,77,79,86,87,88,96,98,102,106,107,108,109,113,114,119,120,121,123,124,130,131,133) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  interactivity=interactivity, compiler=compiler, result=result)\n"
     ]
    }
   ],
   "source": [
    "df = pd.read_csv('resources/slubkin_992016-20160126135704.txt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>STATE_CODE_001</th>\n",
       "      <th>STRUCTURE_NUMBER_008</th>\n",
       "      <th>RECORD_TYPE_005A</th>\n",
       "      <th>ROUTE_PREFIX_005B</th>\n",
       "      <th>SERVICE_LEVEL_005C</th>\n",
       "      <th>ROUTE_NUMBER_005D</th>\n",
       "      <th>DIRECTION_005E</th>\n",
       "      <th>HIGHWAY_DISTRICT_002</th>\n",
       "      <th>COUNTY_CODE_003</th>\n",
       "      <th>PLACE_CODE_004</th>\n",
       "      <th>...</th>\n",
       "      <th>PROJ_NO</th>\n",
       "      <th>PROJ_SUFFIX</th>\n",
       "      <th>NBI_TYPE_OF_IMP</th>\n",
       "      <th>DTL_TYPE_OF_IMP</th>\n",
       "      <th>SPECIAL_CODE</th>\n",
       "      <th>STEP_CODE</th>\n",
       "      <th>STATUS_WITH_10YR_RULE</th>\n",
       "      <th>SUFFICIENCY_ASTERC</th>\n",
       "      <th>SUFFICIENCY_RATING</th>\n",
       "      <th>STATUS_NO_10YR_RULE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>00000000000S702</td>\n",
       "      <td>1</td>\n",
       "      <td>6.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>04007</td>\n",
       "      <td>0.0</td>\n",
       "      <td>9</td>\n",
       "      <td>53.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>100.0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>00000000000S703</td>\n",
       "      <td>1</td>\n",
       "      <td>6.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>04003</td>\n",
       "      <td>0.0</td>\n",
       "      <td>9</td>\n",
       "      <td>53.0</td>\n",
       "      <td>3004.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>100.0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>0000000000M0022</td>\n",
       "      <td>1</td>\n",
       "      <td>8.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>00000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>113.0</td>\n",
       "      <td>27568.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>68.5</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>000000883039900</td>\n",
       "      <td>1</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>00088</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2</td>\n",
       "      <td>59.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>90.2</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>000001014002450</td>\n",
       "      <td>1</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>00101</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2</td>\n",
       "      <td>79.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>59.0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 134 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   STATE_CODE_001 STRUCTURE_NUMBER_008 RECORD_TYPE_005A  ROUTE_PREFIX_005B  \\\n",
       "0               1      00000000000S702                1                6.0   \n",
       "1               1      00000000000S703                1                6.0   \n",
       "2               1      0000000000M0022                1                8.0   \n",
       "3               1      000000883039900                1                4.0   \n",
       "4               1      000001014002450                1                3.0   \n",
       "\n",
       "   SERVICE_LEVEL_005C ROUTE_NUMBER_005D  DIRECTION_005E HIGHWAY_DISTRICT_002  \\\n",
       "0                 1.0             04007             0.0                    9   \n",
       "1                 1.0             04003             0.0                    9   \n",
       "2                 1.0             00000             0.0                    0   \n",
       "3                 1.0             00088             0.0                    2   \n",
       "4                 1.0             00101             0.0                    2   \n",
       "\n",
       "   COUNTY_CODE_003  PLACE_CODE_004         ...          PROJ_NO  PROJ_SUFFIX  \\\n",
       "0             53.0             0.0         ...              NaN          NaN   \n",
       "1             53.0          3004.0         ...              NaN          NaN   \n",
       "2            113.0         27568.0         ...              NaN          NaN   \n",
       "3             59.0             0.0         ...              NaN          NaN   \n",
       "4             79.0             0.0         ...              NaN          NaN   \n",
       "\n",
       "  NBI_TYPE_OF_IMP DTL_TYPE_OF_IMP  SPECIAL_CODE  STEP_CODE  \\\n",
       "0             NaN             NaN           NaN        NaN   \n",
       "1             NaN             NaN           NaN        NaN   \n",
       "2             NaN             NaN           NaN        NaN   \n",
       "3             NaN             NaN           NaN        NaN   \n",
       "4             NaN             NaN           NaN        NaN   \n",
       "\n",
       "   STATUS_WITH_10YR_RULE SUFFICIENCY_ASTERC  SUFFICIENCY_RATING  \\\n",
       "0                      0                NaN               100.0   \n",
       "1                      0                NaN               100.0   \n",
       "2                      0                NaN                68.5   \n",
       "3                      0                NaN                90.2   \n",
       "4                      2                NaN                59.0   \n",
       "\n",
       "   STATUS_NO_10YR_RULE  \n",
       "0                    0  \n",
       "1                    0  \n",
       "2                    0  \n",
       "3                    0  \n",
       "4                    2  \n",
       "\n",
       "[5 rows x 134 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['DECK_STRUCTURE_TYPE_107']"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "[v for v in df.columns if \"107\" in v.lower()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0           2\n",
       "1           2\n",
       "2           1\n",
       "3           1\n",
       "4           1\n",
       "5           1\n",
       "6           6\n",
       "7           6\n",
       "8           1\n",
       "9           8\n",
       "10          2\n",
       "11          1\n",
       "12          2\n",
       "13          N\n",
       "14          1\n",
       "15          8\n",
       "16          1\n",
       "17          1\n",
       "18          1\n",
       "19          N\n",
       "20          8\n",
       "21          8\n",
       "22          8\n",
       "23          8\n",
       "24          1\n",
       "25          1\n",
       "26          8\n",
       "27          N\n",
       "28          3\n",
       "29          3\n",
       "         ... \n",
       "723272    NaN\n",
       "723273    NaN\n",
       "723274    NaN\n",
       "723275    NaN\n",
       "723276    NaN\n",
       "723277    NaN\n",
       "723278    NaN\n",
       "723279    NaN\n",
       "723280    NaN\n",
       "723281    NaN\n",
       "723282    NaN\n",
       "723283    NaN\n",
       "723284    NaN\n",
       "723285    NaN\n",
       "723286    NaN\n",
       "723287    NaN\n",
       "723288    NaN\n",
       "723289    NaN\n",
       "723290    NaN\n",
       "723291    NaN\n",
       "723292    NaN\n",
       "723293    NaN\n",
       "723294    NaN\n",
       "723295    NaN\n",
       "723296    NaN\n",
       "723297    NaN\n",
       "723298    NaN\n",
       "723299    NaN\n",
       "723300    NaN\n",
       "723301    NaN\n",
       "Name: DECK_STRUCTURE_TYPE_107, dtype: object"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.DECK_STRUCTURE_TYPE_107"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Numeric features\n",
    "Try to pull out some features that make sense as predictors of sufficiency."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Average daily traffic for trucks\n",
    "df['TRUCK_ADT'] = df.ADT_029*df.PERCENT_ADT_TRUCK_109/100\n",
    "\n",
    "# Age of bridge\n",
    "df['AGE'] = 2016 - df.YEAR_BUILT_027\n",
    "\n",
    "features = ['AGE',\n",
    "            'ADT_029',\n",
    "            'TRUCK_ADT',\n",
    "            'STRUCTURE_LEN_MT_049',\n",
    "            'TRAFFIC_LANES_ON_028A']\n",
    "\n",
    "label_col = 'SUFFICIENCY_RATING'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Unfortunately, these features (other than age) aren't very strongly correlated with the SR, so a linear model may not work well."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>AGE</th>\n",
       "      <th>ADT_029</th>\n",
       "      <th>TRUCK_ADT</th>\n",
       "      <th>STRUCTURE_LEN_MT_049</th>\n",
       "      <th>TRAFFIC_LANES_ON_028A</th>\n",
       "      <th>SUFFICIENCY_RATING</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AGE</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.021259</td>\n",
       "      <td>-0.024435</td>\n",
       "      <td>-0.047113</td>\n",
       "      <td>-0.067606</td>\n",
       "      <td>-0.500184</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ADT_029</th>\n",
       "      <td>-0.021259</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.722974</td>\n",
       "      <td>0.109050</td>\n",
       "      <td>0.438974</td>\n",
       "      <td>0.013651</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TRUCK_ADT</th>\n",
       "      <td>-0.024435</td>\n",
       "      <td>0.722974</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.082236</td>\n",
       "      <td>0.311347</td>\n",
       "      <td>0.025192</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>STRUCTURE_LEN_MT_049</th>\n",
       "      <td>-0.047113</td>\n",
       "      <td>0.109050</td>\n",
       "      <td>0.082236</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.056318</td>\n",
       "      <td>0.006308</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TRAFFIC_LANES_ON_028A</th>\n",
       "      <td>-0.067606</td>\n",
       "      <td>0.438974</td>\n",
       "      <td>0.311347</td>\n",
       "      <td>0.056318</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.052400</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SUFFICIENCY_RATING</th>\n",
       "      <td>-0.500184</td>\n",
       "      <td>0.013651</td>\n",
       "      <td>0.025192</td>\n",
       "      <td>0.006308</td>\n",
       "      <td>0.052400</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                            AGE   ADT_029  TRUCK_ADT  STRUCTURE_LEN_MT_049  \\\n",
       "AGE                    1.000000 -0.021259  -0.024435             -0.047113   \n",
       "ADT_029               -0.021259  1.000000   0.722974              0.109050   \n",
       "TRUCK_ADT             -0.024435  0.722974   1.000000              0.082236   \n",
       "STRUCTURE_LEN_MT_049  -0.047113  0.109050   0.082236              1.000000   \n",
       "TRAFFIC_LANES_ON_028A -0.067606  0.438974   0.311347              0.056318   \n",
       "SUFFICIENCY_RATING    -0.500184  0.013651   0.025192              0.006308   \n",
       "\n",
       "                       TRAFFIC_LANES_ON_028A  SUFFICIENCY_RATING  \n",
       "AGE                                -0.067606           -0.500184  \n",
       "ADT_029                             0.438974            0.013651  \n",
       "TRUCK_ADT                           0.311347            0.025192  \n",
       "STRUCTURE_LEN_MT_049                0.056318            0.006308  \n",
       "TRAFFIC_LANES_ON_028A               1.000000            0.052400  \n",
       "SUFFICIENCY_RATING                  0.052400            1.000000  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.loc[:,features+[label_col]].corr()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Try fitting a basic linear regression model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.25424854857356638"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.linear_model import LinearRegression\n",
    "\n",
    "df_sub = df[features + [label_col]].dropna(axis=0)\n",
    "\n",
    "X = df_sub[features]\n",
    "y = df_sub[label_col]\n",
    "\n",
    "model = LinearRegression()\n",
    "model.fit(X, y)\n",
    "model.score(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Look at the feature importances."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-----------------------+------------+\n",
      "| Feature               |   Weight   |\n",
      "+-----------------------+------------+\n",
      "| AGE                   | 0.37031184 |\n",
      "| TRAFFIC_LANES_ON_028A | 0.13865022 |\n",
      "| STRUCTURE_LEN_MT_049  | 0.00323628 |\n",
      "| TRUCK_ADT             | 0.00039252 |\n",
      "| ADT_029               | 0.00002600 |\n",
      "+-----------------------+------------+\n"
     ]
    }
   ],
   "source": [
    "import prettytable\n",
    "\n",
    "t = prettytable.PrettyTable()\n",
    "t.field_names = [\"Feature\", \"Weight\"]\n",
    "t.align[\"Feature\"] = \"l\"\n",
    "\n",
    "t.float_format[\"Weight\"] = \"0.8\"\n",
    "\n",
    "for fw in sorted(zip(features, np.abs(model.coef_)), key=lambda x: x[1], reverse=True):\n",
    "  t.add_row(fw)\n",
    "  \n",
    "print(t)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The SR appears to vary somewhat linearly with AGE."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x128a7ce10>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABUwAAAOWCAYAAAA0jdi9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAACr/A/3OJTVxRzFoAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11769bdd8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.plot.hexbin(x=\"AGE\", y=label_col, gridsize=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Regularization\n",
    "Switch to a regularized model to constrain weights."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.linear_model import Ridge\n",
    "\n",
    "df_sub = df[features + [label_col]].dropna(axis=0)\n",
    "\n",
    "X = df_sub[features]\n",
    "y = df_sub[label_col]\n",
    "\n",
    "model = Ridge(alpha=0.2)\n",
    "model.fit(X, y)\n",
    "model.score(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Use cross-validation to choose the optimal hyperparameter `alpha`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.linear_model import RidgeCV\n",
    "\n",
    "df_sub = df[features + [label_col]].dropna(axis=0)\n",
    "\n",
    "X = df_sub[features]\n",
    "y = df_sub[label_col]\n",
    "\n",
    "model = RidgeCV(alphas=[0.3 ** i for i in range(1,5)])\n",
    "model.fit(X, y)\n",
    "print(\"Score: {0} alpha: {1}\".format(model.score(X, y), model.alpha_))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Try a random forest regressor."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%load_ext autoreload"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "%autoreload 2\n",
    "\n",
    "from helpers import compute_test_train_error\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestRegressor\n",
    "\n",
    "\n",
    "df_sub = df[features + [label_col]].dropna(axis=0).reset_index(drop=True)\n",
    "\n",
    "X = df_sub[features]\n",
    "y = df_sub[label_col].as_matrix()\n",
    "\n",
    "models = [Ridge(), RandomForestRegressor()]\n",
    "\n",
    "for m in models:\n",
    "  print(m)\n",
    "  compute_test_train_error(m, X, y)\n",
    "  print()\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The RF is more accurate, but it actually overfitting the data. Let's try bringing in more features and fitting with the Ridge model.\n",
    "\n",
    "\n",
    "For this, we can use Categorical data\n",
    "```\n",
    "Item 107 - Deck Structure Type\n",
    "\n",
    "         Code               Description\n",
    "           1                Concrete Cast-in-Place\n",
    "           2                Concrete Precast Panels\n",
    "           3                Open Grating\n",
    "           4                Closed Grating\n",
    "           5                Steel plate (includes orthotropic)\n",
    "           6                Corrugated Steel\n",
    "           7                Aluminum\n",
    "           8                Wood or Timber\n",
    "           9                Other\n",
    "           N                Not applicable\n",
    "```\n",
    "\n",
    "Categorical data needs to be vectorized, that is mapped to some $n$-dimensional space."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>DECK_STRUCTURE_TYPE_107</th>\n",
       "      <th>integer_labels</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>N</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>6</td>\n",
       "      <td>5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>8</td>\n",
       "      <td>7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>N</td>\n",
       "      <td>9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>8</td>\n",
       "      <td>7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>N</td>\n",
       "      <td>9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>8</td>\n",
       "      <td>7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>8</td>\n",
       "      <td>7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>8</td>\n",
       "      <td>7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>8</td>\n",
       "      <td>7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>8</td>\n",
       "      <td>7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>N</td>\n",
       "      <td>9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613892</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613893</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613894</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613895</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613896</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613897</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613898</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613899</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613900</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613901</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613902</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613903</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613904</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613905</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613906</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613907</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613908</th>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613909</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613910</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613911</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613912</th>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613913</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613914</th>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613915</th>\n",
       "      <td>N</td>\n",
       "      <td>9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613916</th>\n",
       "      <td>N</td>\n",
       "      <td>9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613917</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613918</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613919</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613920</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613921</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>613922 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       DECK_STRUCTURE_TYPE_107  integer_labels    1    2    3    4    5    6  \\\n",
       "0                            2               1  0.0  1.0  0.0  0.0  0.0  0.0   \n",
       "1                            2               1  0.0  1.0  0.0  0.0  0.0  0.0   \n",
       "2                            1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "3                            1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "4                            1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "5                            1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "6                            6               5  0.0  0.0  0.0  0.0  0.0  1.0   \n",
       "7                            6               5  0.0  0.0  0.0  0.0  0.0  1.0   \n",
       "8                            1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "9                            8               7  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "10                           2               1  0.0  1.0  0.0  0.0  0.0  0.0   \n",
       "11                           1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "12                           2               1  0.0  1.0  0.0  0.0  0.0  0.0   \n",
       "13                           N               9  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "14                           1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "15                           8               7  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "16                           1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "17                           1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "18                           1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "19                           N               9  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "20                           8               7  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "21                           8               7  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "22                           8               7  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "23                           8               7  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "24                           1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "25                           1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "26                           8               7  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "27                           N               9  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "28                           3               2  0.0  0.0  1.0  0.0  0.0  0.0   \n",
       "29                           3               2  0.0  0.0  1.0  0.0  0.0  0.0   \n",
       "...                        ...             ...  ...  ...  ...  ...  ...  ...   \n",
       "613892                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613893                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613894                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613895                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613896                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613897                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613898                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613899                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613900                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613901                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613902                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613903                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613904                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613905                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613906                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613907                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613908                       5               4  0.0  0.0  0.0  0.0  1.0  0.0   \n",
       "613909                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613910                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613911                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613912                       5               4  0.0  0.0  0.0  0.0  1.0  0.0   \n",
       "613913                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613914                       5               4  0.0  0.0  0.0  0.0  1.0  0.0   \n",
       "613915                       N               9  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613916                       N               9  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613917                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613918                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613919                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613920                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "613921                       1               0  1.0  0.0  0.0  0.0  0.0  0.0   \n",
       "\n",
       "          7    8    9    N  \n",
       "0       0.0  0.0  0.0  0.0  \n",
       "1       0.0  0.0  0.0  0.0  \n",
       "2       0.0  0.0  0.0  0.0  \n",
       "3       0.0  0.0  0.0  0.0  \n",
       "4       0.0  0.0  0.0  0.0  \n",
       "5       0.0  0.0  0.0  0.0  \n",
       "6       0.0  0.0  0.0  0.0  \n",
       "7       0.0  0.0  0.0  0.0  \n",
       "8       0.0  0.0  0.0  0.0  \n",
       "9       0.0  1.0  0.0  0.0  \n",
       "10      0.0  0.0  0.0  0.0  \n",
       "11      0.0  0.0  0.0  0.0  \n",
       "12      0.0  0.0  0.0  0.0  \n",
       "13      0.0  0.0  0.0  1.0  \n",
       "14      0.0  0.0  0.0  0.0  \n",
       "15      0.0  1.0  0.0  0.0  \n",
       "16      0.0  0.0  0.0  0.0  \n",
       "17      0.0  0.0  0.0  0.0  \n",
       "18      0.0  0.0  0.0  0.0  \n",
       "19      0.0  0.0  0.0  1.0  \n",
       "20      0.0  1.0  0.0  0.0  \n",
       "21      0.0  1.0  0.0  0.0  \n",
       "22      0.0  1.0  0.0  0.0  \n",
       "23      0.0  1.0  0.0  0.0  \n",
       "24      0.0  0.0  0.0  0.0  \n",
       "25      0.0  0.0  0.0  0.0  \n",
       "26      0.0  1.0  0.0  0.0  \n",
       "27      0.0  0.0  0.0  1.0  \n",
       "28      0.0  0.0  0.0  0.0  \n",
       "29      0.0  0.0  0.0  0.0  \n",
       "...     ...  ...  ...  ...  \n",
       "613892  0.0  0.0  0.0  0.0  \n",
       "613893  0.0  0.0  0.0  0.0  \n",
       "613894  0.0  0.0  0.0  0.0  \n",
       "613895  0.0  0.0  0.0  0.0  \n",
       "613896  0.0  0.0  0.0  0.0  \n",
       "613897  0.0  0.0  0.0  0.0  \n",
       "613898  0.0  0.0  0.0  0.0  \n",
       "613899  0.0  0.0  0.0  0.0  \n",
       "613900  0.0  0.0  0.0  0.0  \n",
       "613901  0.0  0.0  0.0  0.0  \n",
       "613902  0.0  0.0  0.0  0.0  \n",
       "613903  0.0  0.0  0.0  0.0  \n",
       "613904  0.0  0.0  0.0  0.0  \n",
       "613905  0.0  0.0  0.0  0.0  \n",
       "613906  0.0  0.0  0.0  0.0  \n",
       "613907  0.0  0.0  0.0  0.0  \n",
       "613908  0.0  0.0  0.0  0.0  \n",
       "613909  0.0  0.0  0.0  0.0  \n",
       "613910  0.0  0.0  0.0  0.0  \n",
       "613911  0.0  0.0  0.0  0.0  \n",
       "613912  0.0  0.0  0.0  0.0  \n",
       "613913  0.0  0.0  0.0  0.0  \n",
       "613914  0.0  0.0  0.0  0.0  \n",
       "613915  0.0  0.0  0.0  1.0  \n",
       "613916  0.0  0.0  0.0  1.0  \n",
       "613917  0.0  0.0  0.0  0.0  \n",
       "613918  0.0  0.0  0.0  0.0  \n",
       "613919  0.0  0.0  0.0  0.0  \n",
       "613920  0.0  0.0  0.0  0.0  \n",
       "613921  0.0  0.0  0.0  0.0  \n",
       "\n",
       "[613922 rows x 12 columns]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.preprocessing import LabelEncoder, OneHotEncoder\n",
    "\n",
    "df_sub = df[[\"DECK_STRUCTURE_TYPE_107\"]].dropna().reset_index(drop=True)\n",
    "\n",
    "le = LabelEncoder()\n",
    "df_sub[\"integer_labels\"] = le.fit_transform(df_sub.DECK_STRUCTURE_TYPE_107)\n",
    "\n",
    "enc = OneHotEncoder()\n",
    "sparse_vectorized = enc.fit_transform(df_sub[[\"integer_labels\"]])\n",
    "df_sub = pd.concat([df_sub, \n",
    "                    pd.DataFrame(sparse_vectorized.todense(), columns=le.classes_)\n",
    "                    ], axis=1)\n",
    "df_sub\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Unfortunately, there is a real deficiency in scikit-learn for vectorizing categorical data. Instead, we have to roll our own Transformers.\n",
    "\n",
    "`SelectTransformer` is used to select a specific column from a `DataFrame`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.base import BaseEstimator, TransformerMixin\n",
    "\n",
    "class SelectTransformer(BaseEstimator, TransformerMixin):\n",
    "  def __init__(self, columns):\n",
    "    self.columns = columns\n",
    "    \n",
    "  def fit(self, X, y=None):\n",
    "    return self\n",
    "  \n",
    "  def transform(self, df):\n",
    "    return df[self.columns].as_matrix()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`CategoricalTransformer` takes a column array and combines two built-in functions to vectorize string categories."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import LabelEncoder, OneHotEncoder\n",
    "\n",
    "  \n",
    "class CategoricalTransformer(BaseEstimator, TransformerMixin):\n",
    "  def __init__(self):\n",
    "    self.le = LabelEncoder()\n",
    "    self.ohe = OneHotEncoder()\n",
    "  \n",
    "  def fit(self, X, y=None):\n",
    "    return self\n",
    "  \n",
    "  def transform(self, X):\n",
    "    labels = self.le.fit_transform(X.ravel())\n",
    "    return self.ohe.fit_transform(labels.reshape(-1,1))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "SelectTransformer([\"DECK_STRUCTURE_TYPE_107\"]).transform(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "CategoricalTransformer().fit_transform(\n",
    "  SelectTransformer([\"DECK_STRUCTURE_TYPE_107\"]).transform(df_sub)\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`Pipeline`s describe both the (*serial*) data transformations and the ultimate Regressor/Classifier that constitute a model.\n",
    "\n",
    "`FeatureUnion`s are used to combine features from disparate sources in *parallel*."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<45403x15 sparse matrix of type '<class 'numpy.float64'>'\n",
       "\twith 265967 stored elements in Compressed Sparse Row format>"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_t = feats.transform(X)\n",
    "X_t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training split 1 of 3\n"
     ]
    }
   ],
   "source": [
    "from sklearn.pipeline import FeatureUnion, Pipeline\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "from sklearn.feature_selection import SelectKBest, mutual_info_regression, SelectFromModel\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "\n",
    "if 1:\n",
    "  df_sub = df\\\n",
    "    .dropna(axis=0, subset=features + [\"DECK_STRUCTURE_TYPE_107\", label_col]).reset_index(drop=True)\n",
    "\n",
    "  X = df_sub\n",
    "  y = df_sub[label_col].as_matrix()\n",
    "\n",
    "  feats = FeatureUnion([\n",
    "            (\"deck_struct\", \n",
    "            Pipeline([\n",
    "              (\"deck\", SelectTransformer([\"DECK_STRUCTURE_TYPE_107\"])),\n",
    "              (\"one_hot\", CategoricalTransformer()),               \n",
    "              ])),\n",
    "            (\"linear\", SelectTransformer(features))\n",
    "          ])\n",
    "\n",
    "  X_t = feats.transform(X)\n",
    "\n",
    "pipe = Pipeline([\n",
    "    ('feature_selection', SelectFromModel(RandomForestRegressor())),\n",
    "    (\"model\", RandomForestRegressor()),\n",
    "    ])\n",
    "# pipe.fit(X_t, y)\n",
    "\n",
    "compute_test_train_error(pipe, X_t, y)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "bins = np.linspace(1800,2016,50)\n",
    "ax = df.YEAR_BUILT_027.hist(bins=bins)\n",
    "ax.set_ylabel('Number of Bridges')\n",
    "ax.set_xlabel('Year Built')\n",
    "df.loc[(df.HISTORY_037 >= 1) & (df.HISTORY_037 <= 3),'YEAR_BUILT_027'].hist(bins=bins,ax=ax)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [conda root]",
   "language": "python",
   "name": "conda-root-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}