{ "metadata": { "name": "", "signature": "sha256:407a2b69c6af15b1c7dcf8065bf72669358f76812d11060e22ac218f3b36a53e" }, "nbformat": 3, "nbformat_minor": 0, "worksheets": [ { "cells": [ { "cell_type": "code", "collapsed": false, "input": [ "import numpy as np\n", "import scipy as sp\n", "import pandas as pd\n", "import sklearn\n", "import seaborn as sns\n", "from matplotlib import pyplot as plt\n", "import bs4\n", "import statsmodels.formula.api as smf\n", "\n", "%matplotlib inline\n", "\n", "#nice defaults for matplotlib\n", "from matplotlib import rcParams\n", "\n", "dark2_colors = [(0.10588235294117647, 0.6196078431372549, 0.4666666666666667),\n", " (0.8509803921568627, 0.37254901960784315, 0.00784313725490196),\n", " (0.4588235294117647, 0.4392156862745098, 0.7019607843137254),\n", " (0.9058823529411765, 0.1607843137254902, 0.5411764705882353),\n", " (0.4, 0.6509803921568628, 0.11764705882352941),\n", " (0.9019607843137255, 0.6705882352941176, 0.00784313725490196),\n", " (0.6509803921568628, 0.4627450980392157, 0.11372549019607843),\n", " (0.4, 0.4, 0.4)]\n", "\n", "rcParams['figure.figsize'] = (10, 6)\n", "rcParams['figure.dpi'] = 150\n", "rcParams['axes.color_cycle'] = dark2_colors\n", "rcParams['lines.linewidth'] = 2\n", "rcParams['axes.grid'] = True\n", "rcParams['axes.facecolor'] = '#eeeeee'\n", "rcParams['font.size'] = 14\n", "rcParams['patch.edgecolor'] = 'none'" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 1 }, { "cell_type": "markdown", "metadata": {}, "source": [ "#Lab 8: Election Prediction\n", "\n" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "##Election scraping\n", "\n", "When was the election?" ] }, { "cell_type": "code", "collapsed": false, "input": [ "#Wikipedia page with table of election dates\n", "theurl = \"http://en.wikipedia.org/wiki/Election_Day_(United_States)\"\n", "\n", "#Grab table from page. Use \"Midterm\" as a unique identifier of the table\n", "#we want. Grab the first returned dataframe.\n", "wikitable = pd.read_html(theurl, match=\"Midterm\", header=0)[0]\n", "#Bingo. But note that the dates are automatically set to 2014.\n", "print wikitable.head()" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "stream", "stream": "stdout", "text": [ " Year Day Details Type\n", "0 2000 2014-11-07 United States elections, 2000 Presidential\n", "1 2001 2014-11-06 United States elections, 2001 Off-year\n", "2 2002 2014-11-05 United States elections, 2002 Midterm\n", "3 2003 2014-11-04 United States elections, 2003 Off-year\n", "4 2004 2014-11-02 United States elections, 2004 Presidential\n" ] } ], "prompt_number": 2 }, { "cell_type": "code", "collapsed": false, "input": [ "def set_year(dateobj, year):\n", " d2 = dateobj.replace(year = year)\n", " return d2\n", "\n", "wikitable['electionday'] = wikitable.apply(lambda r: set_year(r.values[1], r.values[0]), axis=1)\n", "wikitable.head()" ], "language": "python", "metadata": {}, "outputs": [ { "html": [ "
\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
YearDayDetailsTypeelectionday
0 20002014-11-07 United States elections, 2000 Presidential2000-11-07
1 20012014-11-06 United States elections, 2001 Off-year2001-11-06
2 20022014-11-05 United States elections, 2002 Midterm2002-11-05
3 20032014-11-04 United States elections, 2003 Off-year2003-11-04
4 20042014-11-02 United States elections, 2004 Presidential2004-11-02
\n", "
" ], "metadata": {}, "output_type": "pyout", "prompt_number": 3, "text": [ " Year Day Details Type electionday\n", "0 2000 2014-11-07 United States elections, 2000 Presidential 2000-11-07\n", "1 2001 2014-11-06 United States elections, 2001 Off-year 2001-11-06\n", "2 2002 2014-11-05 United States elections, 2002 Midterm 2002-11-05\n", "3 2003 2014-11-04 United States elections, 2003 Off-year 2003-11-04\n", "4 2004 2014-11-02 United States elections, 2004 Presidential 2004-11-02" ] } ], "prompt_number": 3 }, { "cell_type": "code", "collapsed": false, "input": [ "#Grab all data\n", "filedates = [\"2002_2008\", \"2010\", \"2012\"]\n", "pollstercsvs = [pd.read_csv(\"pollster_cleaned_%s.csv\"%fname, parse_dates=['end.date']) \n", " for fname in filedates]\n", "#Paste them together\n", "oldpolls = pd.concat(pollstercsvs)\n", "oldpolls['end.date']\n", "\n", "oldpolls = pd.merge(oldpolls, wikitable, how=\"inner\", left_on=\"year\", right_on=\"Year\")\n", "\n", "#Filter to senate races\n", "oldpolls = oldpolls.ix[oldpolls.race.str.contains(\"Sen\"),:]\n", "\n", "#Compute number of days between poll's end date and election day.\n", "oldpolls['day'] = oldpolls.apply(lambda x: (x['end.date']-x['electionday']).days, axis=1)\n", "#Filter down to polls taken within a year of the election.\n", "oldpolls = oldpolls.ix[(oldpolls.day <= 0) & (oldpolls.day >= -365),:]" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 4 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Deal with independents. Make them dems or republicans depending on who the best competitor is." ] }, { "cell_type": "code", "collapsed": false, "input": [ "def grab_greater(df, focus, champ, chal):\n", " return np.where((df[chal] > df[focus]) & (df[champ] > df[focus]),\n", " df[chal], df[focus])\n", "\n", "oldpolls['real.dem2'] = grab_greater(oldpolls, 'real.dem', 'real.rep', 'real.oth')\n", "oldpolls['real.rep2'] = grab_greater(oldpolls, 'real.rep', 'real.dem', 'real.oth')\n", "\n", "oldpolls['vote.dem2'] = grab_greater(oldpolls, 'vote.dem', 'vote.rep', 'vote.oth')\n", "oldpolls['vote.rep2'] = grab_greater(oldpolls, 'vote.rep', 'vote.dem', 'vote.oth')" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 5 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Now compute differences and bias. Here \"bias\" is a poll's prediction of the Republican's advantage minus the Republican's actual win margin. " ] }, { "cell_type": "code", "collapsed": false, "input": [ "oldpolls['diff'] = oldpolls['vote.rep2']-oldpolls['vote.dem2']\n", "oldpolls['reald'] = oldpolls['real.rep2']-oldpolls['real.dem2']\n", "oldpolls['bias'] = oldpolls['diff']-oldpolls['reald']" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 6 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Note that there are some flipped values." ] }, { "cell_type": "code", "collapsed": false, "input": [ "print oldpolls.ix[oldpolls.bias.abs() >= 50,[\"vote.rep2\", \"vote.dem2\", \"real.rep2\", \"real.dem2\"]]" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "stream", "stream": "stdout", "text": [ " vote.rep2 vote.dem2 real.rep2 real.dem2\n", "4651 32 65 63.36 36.52\n" ] } ], "prompt_number": 7 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Drop these polls." ] }, { "cell_type": "code", "collapsed": false, "input": [ "oldpolls = oldpolls.ix[oldpolls.bias.abs() < 50,:]" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 8 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Now make some exploratory plots. First, plot the number of polls per week per race." ] }, { "cell_type": "code", "collapsed": false, "input": [ "#`value_counts` counts the number of times each value is repeated in a series.\n", "day_table = (oldpolls['day']/7).round().value_counts()\n", "#`nunique` counts the number of unique values in a series.\n", "num_races = oldpolls['race'].nunique()\n", "plt.bar(day_table.index, day_table/num_races)\n", "plt.axhline(y=0.5, color='black', linestyle='--')\n", "plt.axhline(y=1, color='black', linestyle='--')\n", "plt.show()" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAlUAAAFxCAYAAABa010PAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", "text": [ "" ] } ], "prompt_number": 9 }, { "cell_type": "code", "collapsed": false, "input": [ "cutoff = -60\n", "oldpolls = oldpolls.ix[oldpolls.day > cutoff,:]" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 10 }, { "cell_type": "markdown", "metadata": {}, "source": [ "##Which polls should we keep?\n", "\n", "We might want to filter out certain unreliable polls. Explore different groupings of polls (for example, by polling company, party affiliation of the pollster, or sampled population) to see if any stick out in terms of prediction bias." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "###Grouping by pollster.\n", "\n", "Boxplots give a good graphical exploration." ] }, { "cell_type": "code", "collapsed": true, "input": [ "#For this analysis, consider only polls within a week of the election.\n", "tmppolls = oldpolls.ix[oldpolls.day >= -7,]\n", "\n", "#Compute number of polls taken by each pollster, in a format that lets you save this\n", "#back to the original dataframe. Every poll corresponding to a given pollster will\n", "#have the same value for 'npolls' -- `transform` creates these repeats for you.\n", "tmppolls['npolls'] = tmppolls.groupby(\"pollster\")['pollster'].transform(len)\n", "#Only analyze pollsters with more than 2 polls.\n", "tmppolls = tmppolls.ix[tmppolls.npolls >= 2,]\n", "\n", "#Compute mean bias of each pollster. Save this back to the original dataframe.\n", "tmppolls['mean_bias'] = tmppolls.groupby(\"pollster\")['bias'].transform(np.mean)\n", "#Make boxplot grouped by and sorted by mean bias.\n", "tmppolls.boxplot(\"bias\", by=\"mean_bias\", rot=90)\n", "\n", "#To get the right labels on the boxplot, recompute bias, but this time in a way\n", "#that only returns one number per poll. `aggregate` only returns one number per\n", "#pollster group.\n", "pollster_bias = tmppolls.groupby(\"pollster\")['bias'].aggregate(np.mean)\n", "#Sort the pollster biases, then plot their indices to label the boxplot.\n", "pollster_bias.sort()\n", "plt.xticks(np.arange(len(pollster_bias)), pollster_bias.index, fontsize=10)\n", "plt.ylim((-25,25))\n", "plt.show()" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAmUAAAJaCAYAAABuozBfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkqRpwKJMkiRpGrAokyRJmgb+H7XQLBx1pLqfAAAAAElFTkSuQmCC\n", "text": [ "" ] } ], "prompt_number": 11 }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can also explore these biases using a linear regression to explain `bias in terms of the categorical variable `pollster`." ] }, { "cell_type": "code", "collapsed": false, "input": [ "import statsmodels.formula.api as smf\n", "\n", "pollster_anova = smf.ols(formula=\"bias ~ pollster - 1\", data=tmppolls).fit()\n", "pollster_anova.summary()" ], "language": "python", "metadata": {}, "outputs": [ { "html": [ "\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
OLS Regression Results
Dep. Variable: bias R-squared: 0.188
Model: OLS Adj. R-squared: 0.094
Method: Least Squares F-statistic: 2.009
Date: Sun, 26 Oct 2014 Prob (F-statistic): 9.19e-05
Time: 18:37:17 Log-Likelihood: -1539.3
No. Observations: 503 AIC: 3183.
Df Residuals: 451 BIC: 3402.
Df Model: 52
\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
coef std err t P>|t| [95.0% Conf. Int.]
pollster[ARG] -4.0400 3.855 -1.048 0.295 -11.616 3.536
pollster[Angus-Reid] 3.6800 2.438 1.509 0.132 -1.111 8.471
pollster[BlumWeprin] 2.5167 3.148 0.800 0.424 -3.669 8.702
pollster[CNN] 2.2875 2.726 0.839 0.402 -3.069 7.644
pollster[CNN/Time] 2.7400 2.438 1.124 0.262 -2.051 7.531
pollster[Columbus Dispatch] -1.1500 3.855 -0.298 0.766 -8.726 6.426
pollster[ColumbusDispatch] -4.3450 3.855 -1.127 0.260 -11.921 3.231
pollster[Dan Jones & Associates] -0.0500 3.855 -0.013 0.990 -7.626 7.526
pollster[EPIC] -0.2750 2.726 -0.101 0.920 -5.632 5.082
pollster[FOX] 4.4500 2.726 1.633 0.103 -0.907 9.807
pollster[FairleighDickinsonUniv] -3.4450 3.855 -0.894 0.372 -11.021 4.131
pollster[Field] 0.5250 3.855 0.136 0.892 -7.051 8.101
pollster[Gallup] -0.0250 1.724 -0.015 0.988 -3.413 3.363
pollster[Ipsos/Reuters (Web)] 0.6111 1.285 0.476 0.635 -1.914 3.136
pollster[JZ Analytics/Newsmax] -0.4444 1.817 -0.245 0.807 -4.016 3.127
pollster[Marist/McClatchy] 3.8000 2.726 1.394 0.164 -1.557 9.157
pollster[MaristCollege] 4.0933 3.148 1.300 0.194 -2.092 10.279
pollster[MarketingWorkshop] -7.2900 3.855 -1.891 0.059 -14.866 0.286
pollster[Mason-Dixon] 0.7250 2.726 0.266 0.790 -4.632 6.082
pollster[MasonDixon] 4.0093 1.012 3.960 0.000 2.020 5.999
pollster[McLaughlin (R-Mourdock)] 7.2000 3.855 1.868 0.062 -0.376 14.776
pollster[MinneapolistStarTribune] -3.0820 2.438 -1.264 0.207 -7.873 1.709
pollster[Mitchell] 3.6100 3.148 1.147 0.252 -2.576 9.796
pollster[MonmouthUniv] 0.5550 3.855 0.144 0.886 -7.021 8.131
pollster[Muhlenberg] 2.3333 2.226 1.048 0.295 -2.041 6.707
pollster[NBC/WSJ/Marist] 2.6667 3.148 0.847 0.397 -3.519 8.852
pollster[OnPoint] -1.6375 2.726 -0.601 0.548 -6.994 3.719
pollster[POS] 5.4167 3.148 1.721 0.086 -0.769 11.602
pollster[PPP] 1.2230 1.724 0.709 0.478 -2.165 4.611
pollster[PPP (D)] 3.1200 0.995 3.135 0.002 1.164 5.076
pollster[Pulse Opinion Research/Let Freedom Ring (R)] 6.8667 3.148 2.182 0.030 0.681 13.052
pollster[Quinnipiac] -0.1333 3.148 -0.042 0.966 -6.319 6.052
pollster[QuinnipiacUniv] 1.7900 3.148 0.569 0.570 -4.396 7.976
pollster[Rasmussen] 1.9411 0.909 2.136 0.033 0.155 3.727
pollster[Research2000] -3.0276 1.190 -2.545 0.011 -5.366 -0.690
pollster[RutgersUniv] 1.4800 3.855 0.384 0.701 -6.096 9.056
pollster[ScrippsDataCenter] 1.1700 3.855 0.304 0.762 -6.406 8.746
pollster[Selzer] 1.3150 2.726 0.482 0.630 -4.042 6.672
pollster[Siena] 3.3000 3.855 0.856 0.392 -4.276 10.876
pollster[StrategicVision] 3.0444 1.817 1.675 0.095 -0.527 6.616
pollster[SurveyUSA] 0.9203 0.896 1.027 0.305 -0.841 2.682
pollster[Susquehanna] 0.4000 3.148 0.127 0.899 -5.786 6.586
pollster[TalmeyDrake] -7.9300 3.855 -2.057 0.040 -15.506 -0.354
pollster[Tarrance] 2.8050 3.855 0.728 0.467 -4.771 10.381
pollster[The Ohio Poll/University of Cincinnati] 1.5000 3.855 0.389 0.697 -6.076 9.076
pollster[UNH] -3.1320 2.438 -1.285 0.200 -7.923 1.659
pollster[UnivCincinnati] 2.6550 3.855 0.689 0.491 -4.921 10.231
pollster[We Ask America] 4.4000 2.726 1.614 0.107 -0.957 9.757
pollster[Wenzel Strategies (R-Citizens United)] 9.2500 3.855 2.400 0.017 1.674 16.826
pollster[YouGov] 1.8127 0.598 3.029 0.003 0.637 2.989
pollster[YouGov/Polimetrix] 0.3000 1.927 0.156 0.876 -3.488 4.088
pollster[Zogby] 0.9743 0.595 1.638 0.102 -0.195 2.143
\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
Omnibus: 29.210 Durbin-Watson: 1.448
Prob(Omnibus): 0.000 Jarque-Bera (JB): 96.760
Skew: 0.062 Prob(JB): 9.75e-22
Kurtosis: 5.145 Cond. No. 6.48
" ], "metadata": {}, "output_type": "pyout", "prompt_number": 12, "text": [ "\n", "\"\"\"\n", " OLS Regression Results \n", "==============================================================================\n", "Dep. Variable: bias R-squared: 0.188\n", "Model: OLS Adj. R-squared: 0.094\n", "Method: Least Squares F-statistic: 2.009\n", "Date: Sun, 26 Oct 2014 Prob (F-statistic): 9.19e-05\n", "Time: 18:37:17 Log-Likelihood: -1539.3\n", "No. Observations: 503 AIC: 3183.\n", "Df Residuals: 451 BIC: 3402.\n", "Df Model: 52 \n", "=========================================================================================================================\n", " coef std err t P>|t| [95.0% Conf. Int.]\n", "-------------------------------------------------------------------------------------------------------------------------\n", "pollster[ARG] -4.0400 3.855 -1.048 0.295 -11.616 3.536\n", "pollster[Angus-Reid] 3.6800 2.438 1.509 0.132 -1.111 8.471\n", "pollster[BlumWeprin] 2.5167 3.148 0.800 0.424 -3.669 8.702\n", "pollster[CNN] 2.2875 2.726 0.839 0.402 -3.069 7.644\n", "pollster[CNN/Time] 2.7400 2.438 1.124 0.262 -2.051 7.531\n", "pollster[Columbus Dispatch] -1.1500 3.855 -0.298 0.766 -8.726 6.426\n", "pollster[ColumbusDispatch] -4.3450 3.855 -1.127 0.260 -11.921 3.231\n", "pollster[Dan Jones & Associates] -0.0500 3.855 -0.013 0.990 -7.626 7.526\n", "pollster[EPIC] -0.2750 2.726 -0.101 0.920 -5.632 5.082\n", "pollster[FOX] 4.4500 2.726 1.633 0.103 -0.907 9.807\n", "pollster[FairleighDickinsonUniv] -3.4450 3.855 -0.894 0.372 -11.021 4.131\n", "pollster[Field] 0.5250 3.855 0.136 0.892 -7.051 8.101\n", "pollster[Gallup] -0.0250 1.724 -0.015 0.988 -3.413 3.363\n", "pollster[Ipsos/Reuters (Web)] 0.6111 1.285 0.476 0.635 -1.914 3.136\n", "pollster[JZ Analytics/Newsmax] -0.4444 1.817 -0.245 0.807 -4.016 3.127\n", "pollster[Marist/McClatchy] 3.8000 2.726 1.394 0.164 -1.557 9.157\n", "pollster[MaristCollege] 4.0933 3.148 1.300 0.194 -2.092 10.279\n", "pollster[MarketingWorkshop] -7.2900 3.855 -1.891 0.059 -14.866 0.286\n", "pollster[Mason-Dixon] 0.7250 2.726 0.266 0.790 -4.632 6.082\n", "pollster[MasonDixon] 4.0093 1.012 3.960 0.000 2.020 5.999\n", "pollster[McLaughlin (R-Mourdock)] 7.2000 3.855 1.868 0.062 -0.376 14.776\n", "pollster[MinneapolistStarTribune] -3.0820 2.438 -1.264 0.207 -7.873 1.709\n", "pollster[Mitchell] 3.6100 3.148 1.147 0.252 -2.576 9.796\n", "pollster[MonmouthUniv] 0.5550 3.855 0.144 0.886 -7.021 8.131\n", "pollster[Muhlenberg] 2.3333 2.226 1.048 0.295 -2.041 6.707\n", "pollster[NBC/WSJ/Marist] 2.6667 3.148 0.847 0.397 -3.519 8.852\n", "pollster[OnPoint] -1.6375 2.726 -0.601 0.548 -6.994 3.719\n", "pollster[POS] 5.4167 3.148 1.721 0.086 -0.769 11.602\n", "pollster[PPP] 1.2230 1.724 0.709 0.478 -2.165 4.611\n", "pollster[PPP (D)] 3.1200 0.995 3.135 0.002 1.164 5.076\n", "pollster[Pulse Opinion Research/Let Freedom Ring (R)] 6.8667 3.148 2.182 0.030 0.681 13.052\n", "pollster[Quinnipiac] -0.1333 3.148 -0.042 0.966 -6.319 6.052\n", "pollster[QuinnipiacUniv] 1.7900 3.148 0.569 0.570 -4.396 7.976\n", "pollster[Rasmussen] 1.9411 0.909 2.136 0.033 0.155 3.727\n", "pollster[Research2000] -3.0276 1.190 -2.545 0.011 -5.366 -0.690\n", "pollster[RutgersUniv] 1.4800 3.855 0.384 0.701 -6.096 9.056\n", "pollster[ScrippsDataCenter] 1.1700 3.855 0.304 0.762 -6.406 8.746\n", "pollster[Selzer] 1.3150 2.726 0.482 0.630 -4.042 6.672\n", "pollster[Siena] 3.3000 3.855 0.856 0.392 -4.276 10.876\n", "pollster[StrategicVision] 3.0444 1.817 1.675 0.095 -0.527 6.616\n", "pollster[SurveyUSA] 0.9203 0.896 1.027 0.305 -0.841 2.682\n", "pollster[Susquehanna] 0.4000 3.148 0.127 0.899 -5.786 6.586\n", "pollster[TalmeyDrake] -7.9300 3.855 -2.057 0.040 -15.506 -0.354\n", "pollster[Tarrance] 2.8050 3.855 0.728 0.467 -4.771 10.381\n", "pollster[The Ohio Poll/University of Cincinnati] 1.5000 3.855 0.389 0.697 -6.076 9.076\n", "pollster[UNH] -3.1320 2.438 -1.285 0.200 -7.923 1.659\n", "pollster[UnivCincinnati] 2.6550 3.855 0.689 0.491 -4.921 10.231\n", "pollster[We Ask America] 4.4000 2.726 1.614 0.107 -0.957 9.757\n", "pollster[Wenzel Strategies (R-Citizens United)] 9.2500 3.855 2.400 0.017 1.674 16.826\n", "pollster[YouGov] 1.8127 0.598 3.029 0.003 0.637 2.989\n", "pollster[YouGov/Polimetrix] 0.3000 1.927 0.156 0.876 -3.488 4.088\n", "pollster[Zogby] 0.9743 0.595 1.638 0.102 -0.195 2.143\n", "==============================================================================\n", "Omnibus: 29.210 Durbin-Watson: 1.448\n", "Prob(Omnibus): 0.000 Jarque-Bera (JB): 96.760\n", "Skew: 0.062 Prob(JB): 9.75e-22\n", "Kurtosis: 5.145 Cond. No. 6.48\n", "==============================================================================\n", "\"\"\"" ] } ], "prompt_number": 12 }, { "cell_type": "markdown", "metadata": {}, "source": [ "###Group by Party Affiliation" ] }, { "cell_type": "code", "collapsed": false, "input": [ "tmppolls.boxplot(\"bias\", by=\"party\")\n", "plt.show()" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAmUAAAGZCAYAAADIN0bYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", "text": [ "" ] } ], "prompt_number": 13 }, { "cell_type": "code", "collapsed": false, "input": [ "party_anova = smf.ols(formula=\"bias ~ party - 1\", data=tmppolls).fit()\n", "party_anova.summary()\n", "tmppolls.boxplot(\"bias\", by=\"pop\")\n", "plt.show()" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAmUAAAGYCAYAAAADa5V9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", "text": [ "" ] } ], "prompt_number": 14 }, { "cell_type": "markdown", "metadata": {}, "source": [ "###Group by sampled population" ] }, { "cell_type": "code", "collapsed": false, "input": [ "tmppolls.boxplot(\"bias\", by=\"pop\")\n", "plt.show()" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAmUAAAGYCAYAAAADa5V9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", "text": [ "" ] } ], "prompt_number": 15 }, { "cell_type": "code", "collapsed": false, "input": [ "pop_anova = smf.ols(formula=\"bias ~ pop - 1\", data=tmppolls).fit()\n", "pop_anova.summary()" ], "language": "python", "metadata": {}, "outputs": [ { "html": [ "\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
OLS Regression Results
Dep. Variable: bias R-squared: 0.059
Model: OLS Adj. R-squared: 0.053
Method: Least Squares F-statistic: 10.44
Date: Sun, 26 Oct 2014 Prob (F-statistic): 1.13e-06
Time: 18:37:21 Log-Likelihood: -1573.5
No. Observations: 502 AIC: 3153.
Df Residuals: 499 BIC: 3166.
Df Model: 3
\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
coef std err t P>|t| [95.0% Conf. Int.]
pop[General.Population] 5.3100 5.577 0.952 0.341 -5.647 16.267
pop[Likely.Voters] 1.3089 0.262 4.990 0.000 0.794 1.824
pop[Registered.Voters] 1.8731 0.797 2.351 0.019 0.308 3.438
\n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "\n", " \n", "\n", "
Omnibus: 35.247 Durbin-Watson: 1.416
Prob(Omnibus): 0.000 Jarque-Bera (JB): 122.184
Skew: -0.187 Prob(JB): 2.94e-27
Kurtosis: 5.388 Cond. No. 21.3
" ], "metadata": {}, "output_type": "pyout", "prompt_number": 16, "text": [ "\n", "\"\"\"\n", " OLS Regression Results \n", "==============================================================================\n", "Dep. Variable: bias R-squared: 0.059\n", "Model: OLS Adj. R-squared: 0.053\n", "Method: Least Squares F-statistic: 10.44\n", "Date: Sun, 26 Oct 2014 Prob (F-statistic): 1.13e-06\n", "Time: 18:37:21 Log-Likelihood: -1573.5\n", "No. Observations: 502 AIC: 3153.\n", "Df Residuals: 499 BIC: 3166.\n", "Df Model: 3 \n", "===========================================================================================\n", " coef std err t P>|t| [95.0% Conf. Int.]\n", "-------------------------------------------------------------------------------------------\n", "pop[General.Population] 5.3100 5.577 0.952 0.341 -5.647 16.267\n", "pop[Likely.Voters] 1.3089 0.262 4.990 0.000 0.794 1.824\n", "pop[Registered.Voters] 1.8731 0.797 2.351 0.019 0.308 3.438\n", "==============================================================================\n", "Omnibus: 35.247 Durbin-Watson: 1.416\n", "Prob(Omnibus): 0.000 Jarque-Bera (JB): 122.184\n", "Skew: -0.187 Prob(JB): 2.94e-27\n", "Kurtosis: 5.388 Cond. No. 21.3\n", "==============================================================================\n", "\"\"\"" ] } ], "prompt_number": 16 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Subset down to likely voters." ] }, { "cell_type": "code", "collapsed": false, "input": [ "oldpolls = oldpolls.ix[(~oldpolls['pop'].isnull()) & (oldpolls['pop'].str.contains(\"Likely\")),:]" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 17 }, { "cell_type": "code", "collapsed": false, "input": [ "party_anova = smf.ols(formula=\"bias ~ party - 1\", data=tmppolls).fit()\n", "party_anova.summary()\n", "tmppolls.boxplot(\"bias\", by=\"pop\")\n", "plt.show()" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAmUAAAGYCAYAAAADa5V9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XmYXFWZ+PFv6Ihrgo0bozgo4rwDimwuICggBMVRwVEm\nKjoioqKggPpzYVRwA1FBVEQUcaJsRhzFFSSiAsOAggKiwiurCzriEqYBFQT698c5RSpFp9Pp7nSd\ndH0/z9NPV9313KpTt956zzn3zhkdHUWSJEn9tVa/CyBJkiSDMkmSpCYYlEmSJDXAoEySJKkBBmWS\nJEkNMCiTJElqwNx+F0DS9IuIu3om/Qn4KnBgZt6ymve9A/BdYG5m9pZjrOVfCJyXmb9fneXqh4h4\nB7BTZu44xrxDgW0zc8GMF0xSk8yUSbPXHsB6wCOA5wBPBI7qa4l6RMQGwBeB+/e7LJLUb2bKpNlr\naWbeWB//LiIOBz4NvLqPZeo1p+e/JA0sgzJpcPyl+0lEzAHeDLwGeDjwQ+ANmfmTiNgJWALsnJnf\njYgHAj8HPpmZ763No/sA/wE8BPgG8JrMvLl3pxGxPiVDtxNwF/AF4M2ZeRtwbV3sqojYKzM/P8b6\nB9VyPgBYBDwBWJSZn4uI7wM/BZ4F3KfOu9+K9hcRewHvzcxHdm3/+5Tm03dGxCJgBHh0Xf9K4PWZ\neX5d9t7AEcBLKC0NZ9f5N9b5GwOfAraqr+flY74Ty6xd97kH8Dvg7Zl5WkQ8BTgfeESnWTcioh7r\nP2TmH3teo+uBjwCvBDYEzgVemZm/q/NX+B7U12Qf4L+B1wE31dfo+JWUXdI0s/lSmr3uzj5FxIOB\nNwAnds0/BHgTcCCwBXAdcGZEPCAzz6YEQB+LiCHgQ8D/Aod1rf9uYD9gR+DxwD2+xCNibUr/svsB\n21OCj12BI+siT67/t6Y0Y/auv2fdz4HANsCjgKcD3feH2wv4d+B5wC0r2d9YRnu29xrgUmBz4Bzg\nWxHx0DrvMOApwL/UcqxFCUg7Adu3gF8CWwIn122Ndy+7pwG3Ul7/44BTIuIxmfkD4HrghV3LLgS+\n0xuQdTkUOJzyWt4H+HIt18reA4AnAY+rx3YIcExE7DpOuSWtBmbKpNnr6xFxJyU4ux+ls//r4e4s\n2euBgzOzE1S8CriGEuAcSwnYrqAEcv8KPDkz7+za/mGZeWZd9w3Ad2pGrduzKH3anpyZN9Vl9wO+\nERFvBzoBxh8z829jHMN+wMcy87S67suB3/Qs863M/J86/3nj7O/gCbxmAJdm5jvrum+kBHsviojP\n1PJsnZmX1vn/DvwxIrYD1gEeDOybmbcCGRHbA+uPs6/rgP0zcxT4cEQ8G9gX+H/AqZQA6hN12YWU\nLN1YRoFPZ+aptVx7A9dGxGbABhN4Te4EXp6ZfwauqOV+NXDGxF4ySdPBTJk0e70a2IzSpPck4CTg\ngoh4LPBQYBj4QWfhzLwDuBjYuD5fSgnMXgQcnZk/6dn++V2Pf0Q5n0TXtDl1W1d3goHqAmAIeOwE\njmFT4KKuMt4EZNf8UUpGqWOq+4Ou46rB0iXAJpQmzbWB8yLi5oi4GbgRuHfd9ibANTUg67h4Jfv6\ncd3H3c/rMUAJyraLiIdFxOMpzZJfmWC5rwf+XLc1kdcka0DW8aOuckiaIWbKpNnrt5l5bdfzH9Um\nqX2A961gnbks/2Ntc0oWZfuImNMTQNzR9Xio/u/OpI0Cfx1jH0M9/8dzB/ccBND7vDvDNt7+1mLs\npsTu8+Aoyx9DZ/4dXcs9Hfi/nvL8gdJU2Vu2Oxhf7yVD5gK3A2TmzyPicuAFwMOAb47VZ2+cfQ3V\naX8ZY9nu12RF6/a+DpJWMzNl0mBZi3L9sJspHcu37syIiHtROqhnfb4FcACwO7ARtemzy5Zdj59I\nCSau6FnmSmCjiBjumrYN5Qv/asbvbwXws7rtThnn17KsyBUr2d/twLyu7c2hZKA65lD6d3XmD1EC\n08soTbt3Ag/JzGtrwPsHSgf6DSid+jfqacLdgvFt1vP8KZQBFR2nUppPn10fj+fu9yMiNqI0p15G\neT/He00AHhsR9+ua/0RKvzpJM8hMmTR7rRsR69XH9wX2Bh4DnFanHQkcGhE3AFcBb6U0xZ1ag5Hj\ngRMz85sR8TbgIxHxlcz8dV3/fRHxK+A24GOUEZG3lkGCd1sC/AI4sfYhe1Bd9tTMXBoRnR+Gm0fE\n73ua/gA+DnwqIi6hBCvvpozC7ARzc1g+O/WdlezvYmB+7QP3Dcpow+5gBeAZtS/ZNymB6H2AxZl5\nS0QcT+kE/xpKUHs4pYn1F5SA8HrgsxHxH5TA5sV0NRGPYaOIOIpyqZIXUZqaX9A1/9R6zH+r5V2R\nOcBBEXEZZaDBMcBZmXlVRFy9ktcEYD7ldX4/ZfDBCykDOCTNIDNl0uz1ReC39e9nlMsh/GtmXljn\nf4Qy4u9TlD5E6wM7ZOYfKKMdHw28BSAzT6BcjuHYru1/Dvg8cCbwPZbPpI3W9UYpmbZR4EJgMeXO\nAvvU+X+ijPI8hXI5h+Vk5mLKyM9jKf2gfkXpHH97135Gu5Zf2f6uolxe42BKX7G5dZnucn8N2LnO\n3wxYkJkjdf6bgLPqOj+kBGy7ZOZttU/erpSg8WJgf+Do3mPq2ddXKJ3wL6F06n9eZv6263h+TXlv\nvlYvITLethYBH6D0LbuBMjBgpa9J9WvK6NqL6+uzZ2ZeMM7+JK0Gc0ZHV9Z6IEnLq9cp2yEzz13N\n+9me0nn+N/X5XEqT4W6rY98R8Z/AaGbuPd3bnozavHoVsF9mfnuc5a4DDhnrOm8T2Mdedd1HT7qg\nkqaFzZeSWrYb8NSI2JdyDbIDKJ3sLxx3rclr5s4C9fIYO1FaNM7qc3EkzQCDMkktexflOl1LKP3i\nzgeelZm3j7vW5PVeSLafDqRclHfPnlGv062lY5YGms2XkiRJDbCjvyRJUgMMyiRJkhpgUCZJktQA\ngzJJkqQGGJRJkiQ1wKBMkiSpAQZlkiRJDTAokzRrRMQOEXHFGNPfHREv60eZJGmivKK/pFkvMw/p\ndxkkaWUMyiTNNvePiC8CjwVuAl4N/AdweWYeGRF712lrA+sCH8jM4yJiPeDzwIPqdr6Zme+a+eJL\nGlQ2X0qabR4OHJWZWwCnACdR7+8YEfcH9gF2zcwtgRcBH6zrvQq4JjO3Ap4GPDYi5s146SUNLDNl\nkmabn2TmhfXxIuCTwA3AnMy8NSKeAzw3IjYCNgfuX5c9A/hWRPwj8B3gbZl588wWXdIgM1Mmaba5\ns+vxHEqW7O8AEbE+cBnwSOA84B11GTLzYuDRwKeBRwE/jIhtZqzUkgaeQZmk2WaziNi8Pn4N8N/A\nX+rzrYAbM/P9mbkEeC5ARKwVER8A3pmZXwUOBH5G6ZcmSTPCoEzSbDIKXAEcEhGXAc8BXt417yzg\nNxGREXEecBvwO+AxwEeAzSPicuAi4Frg1Bkuv6QBNmd0dHSVV4qIIeB44J8oJ7p9KSe3RcBdwE+B\n/TJz1TcuSZI0gCabKXsOcFdmbkfpk3EYcCRwcGY+ndJHY7fpKaIkSdLsN6mgrPa5eE19+ihgKbBV\nZp5bp50B7Dzl0kmSJA2ISfcpy8w7I2IR8FHgZOoIpuoWYJ2pFU2SJGlwTOk6ZZm5V0Q8DPghcJ+u\nWfMoV9JeoTvvvHN0aGhoKruXJEla08xZ0YxJBWX1xr7rZ+bhwF8p1wW6OCK2z8xzgF2Bs8fbxsjI\nyGR2rWkyPDzM0qVL+10MqS+s/xpk1v/+Gh4eXuG8yWbKvgQsiohzgHsBBwBXAsdHxNrAz+sykiRJ\nmoBJBWWZ+Vdg4RizdphSaSRJkgaUF4+VJElqgEGZJElSAwzKJEmSGmBQJkmS1ACDMkmSpAYYlEmS\nJDXAoEySJKkBBmWSJEkNMCiTJElqgEGZJElSAwzKJEmSGmBQJkmS1ACDMkmSpAYYlEmSJDXAoEyS\nJKkBBmWSJEkNmNvvAkiSpFWzcOFClixZMuP7XbBgAYsXL57x/Q4KM2WSJEkNMFMmSdIaZrLZqk1P\nPwKAy3d/63QWR9PETJkkSVIDDMokSZIaYFAmSZLUAPuUSZI0IC7f/a0MDw+zdOnSfhdFYzBTJkmS\n1ACDMkmSpAYYlEmSJDXAoEySJKkBBmWSJEkNcPSlJEkDwiv6t81MmSRJUgMMyiRJkhpgUCZJktQA\ngzJJkqQGGJRJkiQ1wNGXkiQNCO992TYzZZIkSQ0wUyZpjbRw4UKWLFky4/tdsGABixcvnvH9Spr9\nzJRJkiQ1wEyZpDXSVLJV9qmR1CIzZZIkSQ0wUyZJ0oDw3pdtM1MmSZLUADNlkgaKmQJJrTJTJkmS\n1ACDMkmSpAYYlEmSJDVgUn3KIuJewGeBDYB7A+8DrgAWAXcBPwX2y8zR6SmmJEmaKu992bbJZsr2\nBP6QmU8HngV8AjgSOLhOmwPsNj1FlCRJmv0mO/ryNOBL9fFawN+BLTPz3DrtDGAX4PSpFU+SppeZ\nAkmtmlRQlpm3AkTEPEqA9g7gw12L3AKsM+XSSZIkDYhJX6csIh4JfBn4RGaeGhEf7Jo9D7hpvPXn\nz5/P0NDQZHevaTA8PNzvIkh9Y/3XILP+t2myHf0fBpwFvC4zv1cnXxIR22fmOcCuwNnjbWNkZGQy\nu9Y0sflGg8z6r0Fm/e+v8QLiyWbKDqY0T74rIt5Vpx0AfCwi1gZ+zrI+Z5IkqQHe0aJtk+1TdgAl\nCOu1w5RKI0mSNKC896WkgWKmQFKrvKK/JElSAwzKJEmSGmBQJkmS1AD7lEmSNCC8o0XbzJRJkiQ1\nwEyZpIFipkBSq8yUSZIkNcCgTJIkqQEGZZIkSQ2wT5kkSQPCO1q0zUyZJElSA8yUSRooZgoktcpM\nmSRJUgMMyiRJkhpgUCZJktQA+5RJkjQgvKNF28yUSZIkNcBMmaSBYqZAUqvMlEmSJDXAoEySJKkB\nBmWSJEkNsE+ZJEkDwjtatM1MmSRJUgPMlEkaKGYKJLXKTJkkSVIDDMokSZIaYFAmSZLUAPuUSZI0\nILyjRdvMlEmSJDXATJmkgWKmQFKrzJRJkiQ1wKBMkiSpAQZlkiRJDbBPmSRJA8I7WrTNTJkkSVID\nzJRJGihmCiS1ykyZJElSAwzKJEmSGmBQJkmS1AD7lEmSNCC8o0XbzJRJkiQ1wEyZpIFipkBSq8yU\nSZIkNcCgTJIkqQEGZZIkSQ2YUp+yiHgK8IHM3DEiNgIWAXcBPwX2y8zRqRdRkiRNB+9o0bZJZ8oi\n4i3A8cC966SjgIMz8+nAHGC3qRdPkiRpMEyl+fJq4F8pARjAlpl5bn18BrDzVAomSavDpqcfwfr/\n+bZ+F0OS7mHSQVlmfhm4o2vSnK7HtwDrTHbbkiRJg2Y6r1N2V9fjecBN4y08f/58hoaGpnH3WlXD\nw8P9LoLUN9Z/DTLrf5umMyi7JCK2z8xzgF2Bs8dbeGRkZBp3rVXlxTM16Kz/GmTW//4ZLyCejqCs\nM8LyTcDxEbE28HPgS9OwbUmSNE28o0XbphSUZeb1wFPr46uAHaZeJEmSpMHjvS8lDRQzBZJa5RX9\nJUmSGmBQJkmS1ACDMkmSpAbYp0ySpAHhvS/bZqZMkiSpAWbKJA0UMwWSWmWmTJIkqQEGZZIkSQ0w\nKJMkSWqAfcokSRoQ3tGibWbKJEmSGmCmTNJAMVMgqVVmyiRJkhpgUCZJktQAgzJJkqQG2KdMkqQB\n4R0t2mamTJIkqQFmyiQNFDMFklplpkySJKkBZsr6bP+PfJ1H3Lxuv4sxI26Y92eOOei5/S6GJElN\nMiiTJKkPXnfBaZz3+2v7su9OM/5MedrDNuTYbfaY0X2uiQzK+qxfmSOvaC5J/dWvgKwfBulYp8Kg\nTFLfmCmQZn7QyUz/KJ/pz9qazI7+kvpmkH49D9KxSpocM2WS+s5MgSSZKZMkSWqCQZkkSVIDDMok\nSZIaYFAmSZLUAIMySZKkBhiUSZIkNcCgTJIkqQEGZZIkSQ0wKJMkSWqAQZkkSVIDDMokSZIaYFAm\nSZLUAIMySZKkBhiUSZIkNcCgTJIkqQFz+10ASYNr++s24RE3r8vhPzm/30VZrV7Cdtww78/9Loak\nxhmUSZLUB/4oUS+DMkl9c86jfw7A5bu/dUb3Ozw8zNKlS2dsf5uefkR99NwZ26ekNY9BmSRJfeCP\nEvWyo78kSVIDpjVTFhFrAccCTwBuA/bJzGumcx+SJEmz0XRnynYH1s7MpwJvA46c5u1LkiTNStMd\nlG0LnAmQmT8AnjjN25ckSZqVpjsomw+MdD2/szZpSpIkaRzTPfpyBJjX9XytzLxrrAXnz5/P0NDQ\nNO9eq2J4eLjfRZCA/tTFQdmn2jcoddH6v3LTHZSdTxnzelpEbA38ZEULjoyMrGiWZsBMD4mWxjPT\ndbFf9d/PnMZi/R8s4wWn0x2UfQVYEBGdyxO/Ypq3L0mSNCtNa1CWmaPAa6dzm5IkSYPAK/pLktRH\ny654r0HnyEhJkrRaPe1hG/a7CGsEM2WSJPXBTN/zEpZl5fqxb62cmTJJkqQGGJRJkiQ1wKBMkiSp\nAfYpk9R3jj6TJIMySZoRjj5TCy7f/a3e0aVhBmWS+sbRZ5K0jH3KJEmSGmBQJkmS1ACDMkmSpAYY\nlEmSJDVgzujoaF92vHTp0v7sWACOvtFAs/5rUDnQpf+Gh4fnrGiemTJJkqQGGJRJkiQ1wKBMkiSp\nAQZlkiRJDTAokyRJaoCjLweUo880qBx9pkHn+b+/HH0pSZLUOIMySZKkBhiUSZIkNcCgTJIkqQEG\nZZIkSQ1w9OWAcvSNBpn1X4PK0cf95+hLSZKkxhmUSZIkNcCgTJIkqQEGZZIkSQ0wKJMkSWqAoy8H\nlKPPNKgcfaZB5/m/vxx9KUmS1DiDMkmSpAYYlEmSJDXAoEySJKkBBmWSJEkNcPTlgHL0jQaZ9V+D\nytHH/efoS0mSpMYZlEmSJDXAoEySJKkBBmWSJEkNMCiTJElqgKMvB5SjzzSoHH2mQef5v78cfSlJ\nktQ4gzJJkqQGzJ3sihHxfOCFmblnfb41cDRwB3BWZr5neoooSZI0+00qUxYRHwUOA7rbRT8JvDgz\ntwOeEhGbT0P5JEmSBsJkmy/PB15LDcoiYj5w78y8rs7/NrDz1IsnSZI0GMZtvoyIVwIH9kzeKzO/\nGBE7dE2bD4x0Pb8Z2HBaSihJ0+jy3d/q6DMNLEcft23coCwzTwBOmMB2RoB5Xc/nAzeNt8L8+fMZ\nGhqawKa1ugwPD/e7CFLfWP81yKz/bZp0R/9umTkSEbdHxIbAdcAuwKHjrTMyMjLebK1mZgo0yKz/\nGnTW//4ZLyCeSlA2Wv869gVOBoaAb2fmRVPYtiRJ0kCZdFCWmecA53Q9/wGwzXQUSpIkadB48VhJ\nkqQGeO/LAWWfGg0qR59p0Hn+7y/vfSlJktQ4gzJJkqQGGJRJkiQ1wKBMkiSpAQZlkiRJDXD05YBy\n9I0GmfVfg8rRx/3n6EtJkqTGGZRJkiQ1YFpuSC5JkmbOwoULWbJkyaTXX3fvIya13oIFC1i8ePGk\n96vxmSmTJElqgJkySZLWMFPJVjnQpV0GZZIGiqPPJLXK5ktJkqQGGJRJkiQ1wKBMkiSpAQZlkiRJ\nDTAokyRJaoCjLyUNlMt3f6uXBJDUJDNlkiRJDTAokyRJaoBBmSRJUgMMyiRJkhpgUCZJktQAR19K\nGije+1JSq8yUSZIkNcCgTJIkqQEGZZIkSQ0wKJMkSWqAQZkkSVIDHH0paaB470tJrTJTJkmS1ACD\nMkmSpAYYlEmSJDXAoEySJKkBBmWSJEkNcPSlpIHivS8ltcpMmSRJUgMMyiRJkhpgUCZJktQAgzJJ\nkqQGGJRJkiQ1wNGXkgaK976U1CozZZIkSQ0wKJMkSWrAKjdfRsQ6wEnAPGBt4I2ZeWFEbA0cDdwB\nnJWZ75nWkkqSJM1ik8mUHQQsycwdgL2AT9TpxwEvzsztgKdExObTUkJJkqQBMJmO/h8BbquP7wX8\nNSLmAWtn5nV1+reBnYFLp15ESZKk2W/coCwiXgkc2DN5r8z8UUSsB5wIHACsA4x0LXMzsOF0FlSS\npoP3vpTUqnGDssw8ATihd3pEbAqcCrwpM8+LiPmUPmYd84Gbxtv2/PnzGRoaWvUSa9oMDw/3uwhS\n31j/Ncis/22aTEf/TYDTgD0y83KAzByJiNsjYkPgOmAX4NDxtjMyMjLebK1mXqdJg876r0Hl+b+/\nxguIJ9On7DDKqMuPRQTATZn5fGBf4GRgCPh2Zl40iW1LkiQNpFUOyjJz9xVM/wGwzZRLJEmSNIC8\neKwkSVIDvPelpIHivS8ltcpMmSRJUgMMyiRJkhpgUCZJktQAgzJJkqQGGJRJkiQ1wNGXkgaK976U\n1CozZZIkSQ0wKJMkSWqAQZkkSVIDDMokSZIaYFAmSZLUAEdfShoo3vtSUqvMlEmSJDXAoEySJKkB\nBmWSJEkNMCiTJElqgB39Ja2RFi5cyJIlS2Z8vwsWLGDx4sUzvl9Js5+ZMkmSpAaYKZO0RppKtspL\nYkhqkZkySZKkBhiUSZIkNcCgTJIkqQEGZZIkSQ0wKJMkSWqAQZkkSVIDDMokSZIaYFAmSZLUAIMy\nSZKkBhiUSZIkNcCgTJIkqQEGZZIkSQ0wKJMkSWqAQZkkSVIDDMokSZIaYFAmSZLUAIMySZKkBhiU\nSZIkNcCgTJIkqQEGZZIkSQ0wKJMkSWqAQZkkSVIDDMokSZIaYFAmSZLUAIMySZKkBsxd1RUi4v7A\nKcADgduBl2fmbyNia+Bo4A7grMx8z7SWVJIkaRabTKZsH+CizNweOAl4S51+HPDizNwOeEpEbD5N\nZZQkSZr1Vjkoy8yPAofVpxsASyNiHrB2Zl5Xp38b2Hl6iihJkjT7jdt8GRGvBA7smbxXZv4oIs4G\nHg/sAqwDjHQtczOw4XQWVJIkaTabMzo6OumVIyKAbwJbABdm5uPq9AOAuZl55LSUUpIkaZZb5ebL\niHh7RLysPr0VuCMzbwZuj4gNI2IOJXt27jSWU5IkaVZb5dGXwAnA5yJib2AIeEWdvi9wcp327cy8\naHqKKEmSNPtNqflSkiRJ08OLx0qSJDXAoEySJKkBBmWSJEkNMCiTJElqwGRGX856EXE/4M3AvwGP\nBv4G/A/w3sz8YT/LNpaIWAQMZebLxpi3F/DZrkmjwC2U4zkwM3MGyncXsHNmfncCy24OPCAz/zsi\ndgC+S7nm3V2ruZhaDcZ772vdfG9mPnIq73VEHArslJlPm+DyjwGuAp6dmWeOMf80YK3MfMFKtnN3\nXV2V8mrNEBHXA//YNWkUuAk4D9g/M38zxe3vRa3/E1h2R+DGzPzZVPY5URHxGcp3yit6pu8ELAE2\nycwrx1jvIuDczHzTSrY/o8ezJjFT1qMGZOcCLwD+A9gY2JFyEj8nIp7Ux+KtyGj9W5HfAevVv4cD\nTwXuDXy9XleuJV8B/qk+Ph9Yz4Bs1voCMOP3yM3Ma4AfAC/snRcR9weeDZw4gU1111XNPqPAG1l2\n7lwfWEi5k83npmH7q1L/zwYeNg37nKgVfad8F/gtY392NgS2YmKfnZk+njWGmbJ7egclcNk4M/+v\nTvsV8MaIeBBwMPD8fhVuBebUvxW5KzNv7Hr++4h4O3ABsCnwk9VZuEmYA5CZfwduXMmyWkNl5t8o\nWeh+OAV4V0S8JjPv7Jr+XOB2yp1KJqK1HzWaXiM9587fRcS7gJMiYl69cPqkTKL+z3Rdu8f+MnM0\nIhZTkhbv65n9b8AVmXnpZLcvg7LlRMRawN7AkV0BWbc3UT9EEfE44OPA1sANwCcz86g671Dgn4E/\nAi+lnOSPyswPdO3rHZQL7j6AEhy9PjOvrvPuolT4fYFLM3OXiHgF8BbKPUVHgNPqOt1fKKuis97t\ndZ8bAx8BtqE0b34aeE/9EHaO56+UD95vgLdk5lfrut8HzsvMd9bnjwKuBTbKzGu7dxoR/wB8DNgJ\nuB/wc+CAzDyvbmcD4PiI2Bb4PF1NWhGxPnBUXfcuyi/NN2fmbbUpYB/gLGB/YG1gEXBQZnoxvgaN\n13wTEYdRLky9bWZeO97nrRqt610BfDYzP9S1rQuAL2bmR7qW/yKlLu1A+dXesRA4LTP/HhH3AQ4F\nXgKsS6mL+2fmr3rrambuPYFzwpbAPGAz4EWUz96RlGz8H+ryd58j1Kzb6/87I2IdyvlsN8r58WvA\nmzLzFoCI2Ar4BPAE4BJKXXtaZu7YW/8j4j2U758H1WXfmJkX1mZUgCURcWhmvicitqPU38dTzrUf\nyMyT6nYWUQKeJ1CyeztQztnjlXNbSt39Z+Bb9RhvW8HxnwwcFBGPqVnnjoV1HhExDBwBPA+4L/B1\nyvfV0mk4nh1rOd9D6V70a+CwzFy0gvKuUWy+XN6GwEOBc8aamZl/zMxbIuK+wJmUflmbAq8HDoyI\n/bsWfz6lYm8JfBA4LCL+GSAiXg+8jBKwPRm4Gvhu/RLoeB6wLXBArbCfAN4ObEQJ1l4B/OtkDjIi\nHgG8F7gCyIh4MKWfxG9qeV4L7EcJQruP5856PCcAX4qIx9d5K2s+7XYi5a4P21BS978Gjuvax28o\nTQYH9JR5bcqX4v2A7YE9gF0pX2odT6J8WLet5d8feOYEy6U2zImI/Sh1fEENyCbyeev86j6F8sMB\ngIjYgFIvvtC9k8z8PeUL8gVdy86j1JeT6qTjKJ+xl1ECrbnA1+qPt+Xq6gTL+BzKj6mnAxcC/wV8\nlVJn96Nk7hasyoul1W65bE5tons7cEZm/oXSX3ddYDvgX4Cg/BikBmxnAj+inOtOAd7GGOfKiHg+\npQ68hFIffgx8qc5+Yv2/B/DhiFiPksk9kRLEvAf4eEQ8p2uTe1J+UDyL8sN3vHI+hBKInU35wfAj\nyo+GMc/pmfljIFn+s/NYStB0cp30lfr8OZQf0cGyZs2pHs+NlNfySEr3gcOAz0TErOhKYKZseQ+p\n///UmVD7kH2P5SvoAcCfMvMd9fk1EfFO4J3AMXXaUsovkVFKxXsbpTJeScl47Z+Z36/7eAOlH8sL\nWfaF8OlYbYA2AAAKvElEQVTMvKrO3xLYOzNPr/N+HRFvAjaZ4HE9PCI6afYh4D6UX2Ivrpmwl1Du\nY9ppysl6PO8FPlzXuxHYNzPvAD4YEc8GXgkcNMEydHwV+HJm3lCP7VjgDID6K+pOSpPBzeV+93d7\nFvAI4MmZeVNddz/gGxFxcF1mbj2Gm4GrIuKNlNf8Hp251awXUE6yz8zMn9ZpL2Hln7eOU4F3R8QG\nmflLSoB2bmb+box9nUz5wfS6+nw34A+ZeW79pf9S4F8y8xyAiNiT8iPimZl5Rk9dfeUEyvjHzDy2\nbmtdYJjS2flXwK8i4hnAdav6gmm1mQMcExFH1+f3omSPTqcE3I8BdgcenJlLASLi5cB1Nav/bMp5\n9fW1X+wvakZqvTH29Sjg78CvMvOX9fviSxExlJl/rOfCpZn5l9r15HuZ+fG67rW1peNA4Bt12o+7\nWjJWVs7dKHXz/9V1D6/n9/GcQvm++mB9vhD4n1r2J1B+eGzcGUgWES8FroiIjTPziikezxaUc/1v\nM/PXwKKafZsVXV0Mypa3tP5/YNe0yygRP5QMzOcozQ2P6wp0oGQd146Ie9Xnv+xpNrsZuFdEPIAS\nXJxSmyk77g08tuv59Z0HmfnjiPhbRLybEohtWpf9zgSP6/eUX0hQmv3+nJkjXfM3Bi7paQq9AHhw\nlH50ABfVgKzjYuBxE9x/t+OAF0fEUym/nrYCRiNirXE69M+pZby6E5B1lXGIkj2EcmLpfk9GKCdS\nrTkWAXdQgp+OiXzeAMjMq6OMAFtI+cJYCHxqBfs6HTguIrarIyj/jRLUQfkFvhZlQEBn20sjImt5\nzujZ1oTOCV3b+nNEHAMcG6UrwzeAE2sGT20YpWRnTqM0Ox9CaU15R60L21LOTb/q+QE5Sqk/T6Cc\nV7vPaxcwdgvHKZTs8DW1/n4NOGEF3VM2BnbtqWtzWT4oub5n+fHKuQn37Fd8Mct/D/Y6FTg0Ih5Z\nA6N/A47t2t/NnYAMIDMzIpbWeVdM5Xgy85KI+BrwzYi4htI0uqjnu2GNZfPl8q6mZMm27UzIzNsz\n89raN6rza3su8H1Kqrfztykl7dwJXDr9DrrNYVkgvLBn/Y0pbf4dd3cAjYhnUlLKD6N8GbyQMjJx\nou7sHENmXt8TkAH8hXt2uhzq+d97cpjbNa03zT1msF+bfb5DaRb9JeVL89/H2HevUUo/iF69ZVzR\na641x8uBS4Gju6ZN5PPW7RTghVH6Nm7Ksmag5dQA/uvAHrWpaReWNb+MVd+g1LWhMaZPpIzLderO\nzDdQfph8lPLFeE7tZ6R2/KGeNy+jnLMBTo+IuZT3/BaWf883owQ6F1IyX73fsWN+59ZgfBNKl4wf\nAK8Bflyb9noNUep49z4fT+nW0dHdH2xl5Rwdo1xjfa66y3s1cBHlc7YxpR5/sc5e1c/Oqh4Pmbk7\npSvNSXW5H86Wpn+Dsi41E/RZSifG+WMs8ojOopQK/cuugG0z4G0r61Reo/kbgYd3rXs98H6WZeR6\nvYryS2DfzPwspQl0I6Yv4LgS2LKeaDq2oQSof6jPN+tZ54mULCKUYKj79dpwBfvZBHgasEtmHp6Z\nZ1BGusKyY1nR63clsFFtVuou452UYFqzw39R+mM9PyJ2qdOuZNU+b4uBLagDPzpNNitwCqX/5r8A\nv8jMy+v0ayhfTFt3FqxZ48dSPv+wfF1dpTJGxKMi4jhKc9WHMvPpwH/S1R9ObckyGnwfyvv6Rsp7\n/gDKQKTOez5KGTA1H/gZsFn9MdqxFWP3KdsDeG1mnpWZB1Dq0jzK+fIeRQH+qeuH9rXAglq2MYu+\nknL+FNgiIroDpi3GKmePzmfn+cCZmfnnrv3N6/Shrse3Sd3XWNfFXKXjiYgtIuLozLw0M9+dmVtR\n+kS3dlWESbH58p4OoVzH68LaXHgRsA6lf8n+lDf/JEpa+zMRcQTwSErqdtEE93EU8N6I+D3lg/tm\nSiU8cAXL/xF4akRsSml+fDul0+Z9ehesJ4CHUvq3/H2C5TkFeDfwqYj4MOWL51DKaLDRmvLeKCI+\nBHyGkqnbgvKaQHmN9oqITn+4dzP2B/qmWv4XRcTplA7Ynf5g96Zk7G4BNu4JvqBcsPAXwIm1D8KD\nKJnFU2tTwljHtbJLhWhmPCnKQI1uK8z0ZuZlEXE8pT/P41nFz1tm/m9EnEP54rz7xF6zDjdluRRB\nxxmUYOjNLOvPSWbeWoOmj0XEqyk/UI6gdO7v9FHsrqurek74E6WjM/Uzty7lC/iLK1heDcjMiyPi\nBMqlk06i1IUTowzeuo3ynq9V6+CpwOHA0RHxCcr7uxAY62LDawNHRMTvKK0iO1HO753LS9wCPD4i\nLq77eEOUEcr/SQkSP0Spw2OV+YqIGK+cX6Ccsz8eER+l/EDZntoxf5zvlC9QPhMPpvQ/7uwvI+Ib\nwOdqv985dX/nZWanmXTSxwP8H7BvbQ79PKU/3hNY1vVgjWamrEc9Ye8IHA+8lZIN+h6louydmdtn\nGUb8LEpl+DHlxLuIcrFZWPloxA9T+lZ9om7/cZTOw/+7guUPpTSdXkAZJXMp5cPeufBg9/7+kXJx\nv2261l9Z9u7WejyPqcfzceDozHxX12I/oDSfXkIJyp6dmdfXeUdRTiTnUgK893PP5k6yXAH7tZQP\n2xWUgQLPpjTrbFEXO4aSuj+++7hqtmH3+vxCSjbkqyz70h3rNV+VUaFafQ6n1NvO3zcpgX/v+9P9\n+B2UwPstk/y8La7Tvto17bf0ZKLql8xplJP6KT3beAvwbUrz5/mUZplnZGanmfzuurqqZaxNp8+l\nnFcupTSjnknXl5uadTClafJDlJG5V1EuxfN9StC+G9x9Xn0upT/vZZSuGifVdTs657eTKefNIykZ\nuAOAhVkHe1GyWh8ADqkDQ54D7AxcXstxSGZ+qmubvZ+H8cp5E6XpfgvK+f1ZlPNvx1jfKWS5fts5\ndf7Xevb38rq/symfocspWbWOSR9PzaTtQemb9zNKYHZMbUVa480ZHfU7S+OLco2l7TNzx36XRZqI\nKBf43DgzX9zvsmgw1T6N62fXbbhqxuy+mbl33wqmptl8KWnWqM2dW1D6pS1cyeLS6vRA4Dv1UioX\nU/qTvZRyDTBpTDZfaiJsBtSaYivgk8BJOcZN0KWZkuV2Q/tRmu+vpFx/76A6wEkak82XkiRJDTBT\nJkmS1ACDMkmSpAYYlEmSJDXAoEySJKkBBmWSJEkNMCiTJElqgBePlTTrRMQOlFs1AawP/JByS649\nKffEHKXcGmz/ep/L31BuCbM5cDOwZ2b+cqbLLWmwmSmTNFttTbk35caUmzu/nXLfwqdn5hOAW4FD\n6rIPB87IzM0oN1r+2MwXV9KgMyiTNFt9JzOvqTezPxF4J/C1zFxa538a2Kk+HsnML9THnweeMbNF\nlSSDMkmz1x1dj4eAOfWvYy2WdeG4o2d693NJmhEGZZJmqx0jYr2IWAt4GXAQ8LyIGK7zXwV07o+5\nbkQ8sz5+BfCtmS2qJNnRX9LsdQNwMqW/2FnAMZR+ZOdExL2Ai4F967J/B14WER+s67185osradAZ\nlEmarW7IzF17pp1Q/3rdlZkvnYEySdIK2XwpaTYarX+rsrwk9dWc0VHPRZIkSf1mpkySJKkBBmWS\nJEkNMCiTJElqgEGZJElSAwzKJEmSGvD/AXFEwRQCHRYWAAAAAElFTkSuQmCC\n", "text": [ "" ] } ], "prompt_number": 18 }, { "cell_type": "markdown", "metadata": {}, "source": [ "##A (Not Great) Predictive Model\n", "\n", "The following code implements a prediction strategy that takes all of the polls for given races and generates a point prediction and predicted standard error. The model itself is fairly basic and does not take into account any of the poll aggregation concerns covered in lab. Hint: Set the `plotme` variable to `True` if you'd like to see the model fit for each race." ] }, { "cell_type": "code", "collapsed": false, "input": [ "import patsy\n", "import statsmodels.api as sm\n", "from statsmodels.sandbox.regression.predstd import wls_prediction_std\n", "\n", "#Cutoff for number of polls to perform \"complex\" prediction\n", "#If fewer than minN polls, just average all polls together\n", "#If more, use a spline fit to try to detect a trend.\n", "minN = 15\n", "#Plot the poll history, model fit, and error bar for each race, or not.\n", "plotme= False\n", "results = []\n", "\n", "#Group the polls by race.\n", "oldpolls_groups = oldpolls.groupby(\"race\")\n", "\n", "#Step through, group by group. For each loop iteration:\n", "#race is assigned the name of the current group's race\n", "#oldpolls_small is assigned the dataframe corresponding to that group.\n", "for race, oldpolls_small in oldpolls_groups:\n", " #Check whether to do the simple or spline fit\n", " do_spline = oldpolls_small.shape[0] > minN\n", " \n", " #Simple fit first\n", " if not do_spline:\n", " data = oldpolls_small['diff']\n", " #Estimate is mean of all polls.\n", " #Note that this ignores sample sizes, pollster quality, etc.\n", " pred = np.mean(data)\n", " #Standard error is the standard deviation of the sample mean\n", " pred_se = np.sqrt(np.var(data)*(1/float(oldpolls_small.shape[0])))\n", " \n", " #More complex fit does a spline regression\n", " #Despite being \"fancier\", this approach also ignores samples sizes, pollster quality, etc.\n", " else:\n", " #Build a bunch of smooth functions of time. Throw these into a regression.\n", " spline_mat = patsy.dmatrix(\"cr(day-1, df=4)\", data=oldpolls_small)\n", " pollfit = sm.OLS(oldpolls_small['diff'], spline_mat).fit()\n", "\n", " #Use the same function generator to create a representation of election day.\n", " #We will throw this into the regression model to get an estimate for election day.\n", " predmat = patsy.build_design_matrices([spline_mat.design_info.builder],\n", " pd.DataFrame({'day': [0]}))[0]\n", "\n", " #Get the prediction\n", " pred = pollfit.predict(predmat)[0]\n", " #Predicted standard error formula for regressions, when you expect the future datapoint to have\n", " #no sampling noise.\n", " pred_se = np.sqrt(np.dot(predmat, np.dot(pollfit.cov_params(), predmat.T)).diagonal())[0]\n", " \n", " #Compare to the true result\n", " result = oldpolls_small['reald'].iloc[0]\n", " \n", " #Construct a 99% predictive interval. Check whether it contains the true result.\n", " pred_interval = pred+pred_se*np.array([-2.54, 2.54])\n", " covers = result > pred_interval[0] and result < pred_interval[1]\n", "\n", " #Optional plotting functionality\n", " if plotme:\n", " #Plot the polling data.\n", " plt.scatter(oldpolls_small['day'], oldpolls_small['diff'])\n", " #Draw a line representing the outcome.\n", " plt.axhline(result, color='black', linestyle='--')\n", " \n", " #Draw the model fit.\n", " if do_spline:\n", " order = np.array(oldpolls_small['day'].values.argsort())\n", " plt.plot(oldpolls_small['day'].values[order], np.array(pollfit.predict())[order], color='orange')\n", " else:\n", " plt.plot([oldpolls['day'].min(), oldpolls['day'].max()], [pred, pred], color='orange')\n", " \n", " #Plot the model prediction\n", " plt.scatter(0, pred)\n", " #Draw the 99% confidence interval\n", " plt.plot([0,0], pred_interval)\n", " plt.title(race)\n", " plt.show()\n", " \n", " #Save all of the results of the computation.\n", " results.append((race, oldpolls_small['year'].iloc[0], pred, pred_se, covers, result, do_spline))\n", " \n", "#Turn results into a dataframe for further analysis.\n", "results_df = pd.DataFrame(results, columns=[\"Race\", \"Year\", \"Pred\", \"PredSE\", \"Covers99\", \"Result\", \"Spline\"])\n", "results_df.ix[results_df['PredSE']<0.001, 'PredSE'] = np.median(results_df['PredSE'])" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 19 }, { "cell_type": "markdown", "metadata": {}, "source": [ "##Checking Historical Performance\n", "\n", "Plot the performance of the predictions against the true outcomes. Break these out by year.\n", "\n", "First, look at predicted versus actual **vote share**. Green points/error bars have 99% error bars that contain the actual result; orange points/error bars do not have 99% intervals that contain the result. If the model were **well-calibrated**, each year would have 99% **coverage** (percentage of predictions with 99% intervals that cover the truth). The actual coverage is in the title of each plot." ] }, { "cell_type": "code", "collapsed": false, "input": [ "results_year = results_df.groupby(\"Year\")\n", "\n", "fignum = 0\n", "for yr, year_df in results_year:\n", " covers = year_df['Covers99']\n", " err99 = year_df['PredSE']*2.54\n", " \n", " #Colors for predictions that cover or miss the truth.\n", " cover_color = dark2_colors[0]\n", " miss_color = \"darkorange\"\n", " \n", " fignum += 1 \n", " plt.subplot(2, 3, fignum)\n", " #Plot predictions that cover the truth.\n", " plt.errorbar(year_df.ix[covers, 'Result'],\n", " year_df.ix[covers, 'Pred'],\n", " yerr=err99[covers], fmt='o', c=cover_color, ecolor=cover_color)\n", " #Plot predictions that miss.\n", " plt.errorbar(year_df.ix[~covers, 'Result'],\n", " year_df.ix[~covers, 'Pred'],\n", " yerr=err99[~covers], fmt='o', c=miss_color, ecolor=miss_color)\n", " \n", " plt.xlim((-60,60))\n", " plt.ylim((-60,60))\n", " plt.plot([-60,60], [-60,60], linewidth=1, color='black')\n", " \n", " plt.title(\"%d: %d%%\"%(yr, (np.mean(covers)*100).round()))\n", " plt.xlabel(\"Actual\")\n", " plt.ylabel(\"Predicted + Interval\")\n", " \n", "plt.tight_layout()\n", "plt.show()" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAsAAAAGfCAYAAABRH2RQAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n", "text": [ "" ] } ], "prompt_number": 20 }, { "cell_type": "markdown", "metadata": {}, "source": [ "The model is not very well-calibrated since the 99% prediction intervals contain the truth far less than 99% of the time. However, the predictions seem to track well with the truth, suggesting that our current model is **overconfident** and is giving predictive intervals that are too short. If we made the predictive intervals larger, then they would contain the truth closer to 99% of time time. But these intervals (ostensibly) take into account all of the variation *between* polls. Where is the extra variation coming from? That's what you should think about on your homework." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can do the same sort of performance check with the binary \"Dem or Republican\" outcome. Compute the predictied probability of a Republican win using the normal approximation to our predictive distribution. Look at the probability that a vote percentage difference drawn from this distribution is greater than zero.\n", "\n", "Since we saw that our predictive distributions were too narrow (resulting in low coverage) in the last part, try adding some additional variance to the predictive distribution." ] }, { "cell_type": "code", "collapsed": false, "input": [ "from scipy.stats import norm\n", "\n", "#`norm.cdf` returns the probability that a normal random variable is less than\n", "#a particular value.\n", "#Here we add 2 to the predicted distribution's standard deviation.\n", "#What might this extra variation represent?\n", "results_df['prob_rep_win'] = 1-norm.cdf(0, loc=results_df['Pred'], scale=results_df['PredSE']+2)" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 21 }, { "cell_type": "markdown", "metadata": {}, "source": [ "To see whether these probabilities look better calibrated, group the data by the predicted probability of a Republican win. See whether the empirical win percentage in each group seems to track with the predicted win percentage." ] }, { "cell_type": "code", "collapsed": false, "input": [ "results_df['pred_group'] = pd.cut(results_df['prob_rep_win'], 10)\n", "pred_groups = results_df.groupby(\"pred_group\")\n", "pred_prob = pred_groups['prob_rep_win'].aggregate(np.mean)\n", "real_prob = pred_groups['Result'].aggregate(lambda x: np.mean(x > 0))\n", "\n", "plt.scatter(pred_prob, real_prob)\n", "plt.xlim(0,1)\n", "plt.ylim(0,1)\n", "plt.plot([0,1], [0,1], linewidth=1, color='black')\n", "\n", "plt.title(\"Republican Win Probability\")\n", "plt.xlabel(\"Actual\")\n", "plt.ylabel(\"Predicted\")\n", "plt.gca().set_aspect('equal')" ], "language": "python", "metadata": {}, "outputs": [ { "metadata": {}, "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAGJCAYAAACO1pQwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", "text": [ "" ] } ], "prompt_number": 22 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Adding some variation does seem to help in predicting win percentages. There are still some real issues for the close races near 50%. Those are also the most interesting ones for a forecaster's perspective, so this is a problem. Hopefully, you can do better on the homework!" ] } ], "metadata": {} } ] }