{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing the 2016 General Election Polls"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from __future__ import print_function\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from ipywidgets import VBox, HBox\n",
    "import os\n",
    "\n",
    "codes = pd.read_csv(os.path.abspath('../data_files/state_codes.csv'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "try:\n",
    "    from pollster import Pollster\n",
    "except ImportError:\n",
    "    print('Pollster not found. Installing Pollster..')\n",
    "    try:\n",
    "        import subprocess\n",
    "        subprocess.check_call(['pip', 'install', 'pollster==0.1.6'])\n",
    "    except:\n",
    "        print(\"The pip installation failed. Please manually install Pollster and re-run this notebook.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def get_candidate_data(question):\n",
    "    clinton, trump, undecided, other = 0., 0., 0., 0.\n",
    "    for candidate in question['subpopulations'][0]['responses']:\n",
    "        if candidate['last_name'] == 'Clinton':\n",
    "            clinton = candidate['value']\n",
    "        elif candidate['last_name'] == 'Trump':\n",
    "            trump = candidate['value']\n",
    "        elif candidate['choice'] == 'Undecided':\n",
    "            undecided = candidate['value']\n",
    "        else:\n",
    "            other = candidate['value']\n",
    "    return clinton, trump, other, undecided\n",
    "\n",
    "def get_row(question, partisan='Nonpartisan', end_date='2016-06-21'):\n",
    "    # if question['topic'] != '2016-president':\n",
    "    if ('2016' in question['topic']) and ('Presidential' in question['topic']):\n",
    "        hillary, donald, other, undecided = get_candidate_data(question)\n",
    "        return [{'Name': question['name'], 'Partisan': partisan, 'State': question['state'],\n",
    "                 'Date': np.datetime64(end_date), 'Trump': donald, 'Clinton': hillary, 'Other': other,\n",
    "                 'Undecided': undecided}]\n",
    "    else:\n",
    "        return\n",
    "\n",
    "def analyze_polls(polls):\n",
    "    global data\n",
    "    for poll in polls:\n",
    "        for question in poll.questions:\n",
    "            resp = get_row(question, partisan=poll.partisan, end_date=poll.end_date)\n",
    "            if resp is not None:\n",
    "                data = data.append(resp)\n",
    "    return"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "try:\n",
    "    from pollster import Pollster\n",
    "    pollster = Pollster()\n",
    "    \n",
    "    # Getting data from Pollster. This might take a second.\n",
    "    raw_data = pollster.charts(topic='2016-president')\n",
    "    \n",
    "    data = pd.DataFrame(columns=['Name', 'Partisan', 'State', 'Date', 'Trump', 'Clinton', 'Other',\n",
    "                     'Undecided'])\n",
    "    \n",
    "    for i in raw_data:\n",
    "        analyze_polls(i.polls())\n",
    "except:\n",
    "    raise ValueError('Please install Pollster and run the functions above')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def get_state_party(code):\n",
    "    state = codes[codes['FIPS']==code]['USPS'].values[0]\n",
    "    if data[data['State']==state].shape[0] == 0:\n",
    "        return None\n",
    "    polls = data[(data['State']==state) & (data['Trump'] > 0.) & (data['Clinton'] > 0.)].sort_values(by='Date')\n",
    "    if polls.shape[0] == 0:\n",
    "        return None\n",
    "    if (polls.tail(1)['Trump'] > polls.tail(1)['Clinton']).values[0]:\n",
    "        return 'Republican'\n",
    "    else:\n",
    "        return 'Democrat'\n",
    "\n",
    "def get_color_data():\n",
    "    color_data = {}\n",
    "    for i in codes['FIPS']:\n",
    "        color_data[i] = get_state_party(i)\n",
    "    return color_data\n",
    "\n",
    "def get_state_data(code):\n",
    "    state = codes[codes['FIPS']==code]['USPS'].values[0]\n",
    "    if data[data['State']==state].shape[0] == 0:\n",
    "        return None\n",
    "    polls = data[(data['State']==state) & (data['Trump'] > 0.) & (data['Clinton'] > 0.)].sort_values(by='Date')\n",
    "    return polls"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from bqplot import *\n",
    "from ipywidgets import Layout"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "dt_x = DateScale()\n",
    "sc_y = LinearScale()\n",
    "\n",
    "time_series = Lines(scales={'x': dt_x, 'y': sc_y}, colors=['#E91D0E', '#2aa1ec'], marker='circle')\n",
    "\n",
    "ax_x = Axis(scale=dt_x, label='Date')\n",
    "ax_y = Axis(scale=sc_y, orientation='vertical', label='Percentage')\n",
    "\n",
    "ts_fig = Figure(marks=[time_series], axes=[ax_x, ax_y], title='General Election - State Polls', \n",
    "                layout=Layout(min_width='650px', min_height='400px'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sc_geo = AlbersUSA()\n",
    "sc_c1 = OrdinalColorScale(domain=['Democrat', 'Republican'], colors=['#2aa1ec', '#E91D0E'])\n",
    "\n",
    "color_data = get_color_data()\n",
    "\n",
    "map_styles = {'color': color_data,\n",
    "              'scales': {'projection': sc_geo, 'color': sc_c1}, 'colors': {'default_color': 'Grey'}}\n",
    "\n",
    "axis = ColorAxis(scale=sc_c1)\n",
    "\n",
    "states_map = Map(map_data=topo_load('map_data/USStatesMap.json'), tooltip=ts_fig, **map_styles)\n",
    "map_fig = Figure(marks=[states_map], axes=[axis],title='General Election Polls - State Wise')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def hover_callback(name, value):\n",
    "    polls = get_state_data(value['data']['id'])\n",
    "    if polls is None or polls.shape[0] == 0:\n",
    "        time_series.y = [0.]\n",
    "        return\n",
    "    time_series.x, time_series.y = polls['Date'].values.astype(np.datetime64), [polls['Trump'].values, polls['Clinton'].values]\n",
    "    ts_fig.title = str(codes[codes['FIPS']==value['data']['id']]['Name'].values[0]) + ' Polls - Presidential Election'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "states_map.on_hover(hover_callback)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "national = data[(data['State']=='US') & (data['Trump'] > 0.) & (data['Clinton'] > 0.)].sort_values(by='Date')\n",
    "\n",
    "dt_x = DateScale()\n",
    "sc_y = LinearScale()\n",
    "\n",
    "clinton_scatter = Scatter(x=national['Date'].values.astype(np.datetime64), y=national['Clinton'],\n",
    "                          scales={'x': dt_x, 'y': sc_y}, \n",
    "                          colors=['#2aa1ec'])\n",
    "\n",
    "trump_scatter = Scatter(x=national['Date'].values.astype(np.datetime64), y=national['Trump'],\n",
    "                        scales={'x': dt_x, 'y': sc_y},\n",
    "                        colors=['#E91D0E'])\n",
    "\n",
    "ax_x = Axis(scale=dt_x, label='Date', tick_format='%b-%Y', num_ticks=8)\n",
    "ax_y = Axis(scale=sc_y, orientation='vertical', label='Percentage')\n",
    "\n",
    "scat_fig = Figure(marks=[clinton_scatter, trump_scatter], axes=[ax_x, ax_y], title='General Election - National Polls')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Hover on the map to visualize the poll data for that state."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "VBox([map_fig, scat_fig])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing the County Results of the 2008 Elections"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "county_data = pd.read_csv(os.path.abspath('../data_files/2008-election-results.csv'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "winner = np.array(['McCain'] * county_data.shape[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "winner[(county_data['Obama'] > county_data['McCain']).values] = 'Obama'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "sc_geo_county = AlbersUSA()\n",
    "sc_c1_county = OrdinalColorScale(domain=['McCain', 'Obama'], colors=['Red', 'DeepSkyBlue'])\n",
    "\n",
    "color_data_county = dict(zip(county_data['FIPS'].values.astype(int), list(winner)))\n",
    "\n",
    "map_styles_county = {'color': color_data_county,\n",
    "              'scales': {'projection': sc_geo_county, 'color': sc_c1_county}, 'colors': {'default_color': 'Grey'}}\n",
    "\n",
    "axis_county = ColorAxis(scale=sc_c1_county)\n",
    "\n",
    "county_map = Map(map_data=topo_load('map_data/USCountiesMap.json'), **map_styles_county)\n",
    "county_fig = Figure(marks=[county_map], axes=[axis_county],title='US Elections 2008 - Example',\n",
    "                    layout=Layout(min_width='800px', min_height='550px'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "names_sc = OrdinalScale(domain=['Obama', 'McCain'])\n",
    "vote_sc_y = LinearScale(min=0, max=100.)\n",
    "\n",
    "names_ax = Axis(scale=names_sc, label='Candidate')\n",
    "vote_ax = Axis(scale=vote_sc_y, orientation='vertical', label='Percentage')\n",
    "\n",
    "vote_bars = Bars(scales={'x': names_sc, 'y': vote_sc_y}, colors=['#2aa1ec', '#E91D0E'])\n",
    "\n",
    "bar_fig = Figure(marks=[vote_bars], axes=[names_ax, vote_ax], title='Vote Margin',\n",
    "                 layout=Layout(min_width='600px', min_height='400px'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def county_hover(name, value):\n",
    "    if (county_data['FIPS'] == value['data']['id']).sum() == 0:\n",
    "        bar_fig.title = ''\n",
    "        vote_bars.y = [0., 0.]\n",
    "        return\n",
    "    votes = county_data[county_data['FIPS'] == value['data']['id']]\n",
    "    dem_vote = float(votes['Obama %'].values[0])\n",
    "    rep_vote = float(votes['McCain %'].values[0])\n",
    "    vote_bars.x, vote_bars.y = ['Obama', 'McCain'], [dem_vote, rep_vote]\n",
    "    bar_fig.title = 'Vote % - ' + value['data']['name']\n",
    "    \n",
    "county_map.on_hover(county_hover)\n",
    "county_map.tooltip = bar_fig"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Hover on the map to visualize the voting percentage for each candidate in that county"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "county_fig"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}