saulshanabrook/ideal point.ipynb

## ideal point.ipynb
{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-11T00:07:24.632250",
     "start_time": "2016-12-11T00:07:24.605803"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-11T00:07:25.101393",
     "start_time": "2016-12-11T00:07:24.794314"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [2] used 0.230 MiB RAM in 0.002s, peaked 0.000 MiB above current, total RAM usage 42.523 MiB\n"
     ]
    }
   ],
   "source": [
    "from ipython_memwatcher import MemWatcher\n",
    "mw = MemWatcher()\n",
    "mw.start_watching_memory()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Ideal Point Estimation Testing\n",
    "\n",
    "We want to test our gradient based method of ideal point estimation, against the PyMC3 model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-11T00:07:28.938723",
     "start_time": "2016-12-11T00:07:27.071781"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [3] used 79.211 MiB RAM in 1.753s, peaked 0.000 MiB above current, total RAM usage 121.734 MiB\n"
     ]
    }
   ],
   "source": [
    "import ideal_point.ideal_point as ip"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Generate Test Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-10T03:51:58.380436",
     "start_time": "2016-12-10T03:51:58.240842"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df = ip.test_data(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-10T02:21:44.873097",
     "start_time": "2016-12-10T02:21:44.720807"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-11T00:25:32.871436",
     "start_time": "2016-12-11T00:25:32.381168"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [18] used 9.652 MiB RAM in 0.356s, peaked 0.000 MiB above current, total RAM usage 4340.082 MiB\n"
     ]
    }
   ],
   "source": [
    "from altair import *\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-11T00:25:31.130435",
     "start_time": "2016-12-11T00:25:30.974587"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [17] used 1.191 MiB RAM in 0.016s, peaked 0.000 MiB above current, total RAM usage 4330.430 MiB\n"
     ]
    }
   ],
   "source": [
    "def plot_params(params):\n",
    "    rows = []\n",
    "    for key, value in params.items():\n",
    "        for (i, v) in enumerate(value):\n",
    "            rows.append({\n",
    "                'i': i,\n",
    "                'param': key,\n",
    "                'value': v\n",
    "            })\n",
    "    df = pd.DataFrame(rows)\n",
    "    return Chart(df).mark_circle().encode(\n",
    "        column='param:N',\n",
    "        x='i:O',\n",
    "        y='value:Q'\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### PyMC3 ADVI"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-10T02:15:21.771406",
     "start_time": "2016-12-10T02:15:17.032349"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "pymc_params = ip.advi_params(ip.create_model(df)).means"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-10T02:15:22.785465",
     "start_time": "2016-12-10T02:15:22.623627"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "plot_params(pymc_params)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The `vote_ideology` should match with the `legislator_ideology`. The first vote should be close ot the first legislator, and the last vote should be close to the last legislator (in ideology)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### PyMC3 Sampling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-10T02:23:30.053375",
     "start_time": "2016-12-10T02:22:17.667228"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from pymc3 import find_MAP\n",
    "from pymc3 import NUTS, sample\n",
    "\n",
    "model=ip.create_model(df)\n",
    "with model:\n",
    "    trace = sample(2000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-10T02:23:30.216229",
     "start_time": "2016-12-10T02:23:30.056288"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from pymc3 import forestplot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-10T02:23:31.060811",
     "start_time": "2016-12-10T02:23:30.219276"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "forestplot(trace)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Same problem here..."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Manual Gradient"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-11T00:14:14.428591",
     "start_time": "2016-12-11T00:14:14.113074"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [13] used 0.238 MiB RAM in 0.002s, peaked 0.000 MiB above current, total RAM usage 5992.418 MiB\n"
     ]
    }
   ],
   "source": [
    "from ideal_point.gradient import Gradient \n",
    "from scipy.special import expit as logistic"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-11T00:14:16.315575",
     "start_time": "2016-12-11T00:14:16.172393"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The line_profiler extension is already loaded. To reload it, use:\n",
      "  %reload_ext line_profiler\n",
      "In [14] used 0.000 MiB RAM in 0.003s, peaked 0.000 MiB above current, total RAM usage 5992.418 MiB\n"
     ]
    }
   ],
   "source": [
    "%load_ext line_profiler\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-11T00:25:47.316392",
     "start_time": "2016-12-11T00:25:47.009489"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [20] used -135.172 MiB RAM in 0.170s, peaked 135.406 MiB above current, total RAM usage 4213.129 MiB\n"
     ]
    }
   ],
   "source": [
    "g = Gradient(ip.test_data(30))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-11T00:25:58.036278",
     "start_time": "2016-12-11T00:25:57.826675"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-720.30101696\n",
      "-720.30101696\n",
      "-478.285431624\n",
      "-441.825295499\n",
      "-420.980500451\n",
      "-385.627159114\n",
      "-324.802623769\n",
      "-252.811290555\n",
      "-200.873256169\n",
      "-170.464783023\n",
      "\n",
      "In [21] used 0.934 MiB RAM in 0.075s, peaked 0.000 MiB above current, total RAM usage 4214.062 MiB\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Widget Javascript not detected.  It may not be installed properly. Did you enable the widgetsnbextension? If not, then run \"jupyter nbextension enable --py --sys-prefix widgetsnbextension\"\n"
     ]
    }
   ],
   "source": [
    "g.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-12-11T00:26:01.072789",
     "start_time": "2016-12-11T00:26:00.916794"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div class=\"vega-embed\" id=\"c113c9cd-ea77-4159-9de3-17dfd10cf854\"></div>\n",
       "\n",
       "<style>\n",
       ".vega-embed svg, .vega-embed canvas {\n",
       "  border: 1px dotted gray;\n",
       "}\n",
       "\n",
       ".vega-embed .vega-actions a {\n",
       "  margin-right: 6px;\n",
       "}\n",
       "</style>\n"
      ]
     },
     "metadata": {
      "jupyter-vega": "#c113c9cd-ea77-4159-9de3-17dfd10cf854"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "application/javascript": [
       "var spec = {\"mark\": \"circle\", \"encoding\": {\"x\": {\"field\": \"i\", \"type\": \"ordinal\"}, \"column\": {\"field\": \"param\", \"type\": \"nominal\"}, \"y\": {\"field\": \"value\", \"type\": \"quantitative\"}}, \"config\": {\"cell\": {\"width\": 500, \"height\": 350}}, \"data\": {\"values\": [{\"i\": 0, \"param\": \"vote_ideologies\", \"value\": 0.7238689188660034}, {\"i\": 1, \"param\": \"vote_ideologies\", \"value\": 0.9864073981522212}, {\"i\": 2, \"param\": \"vote_ideologies\", \"value\": 1.0468387844806417}, {\"i\": 3, \"param\": \"vote_ideologies\", \"value\": 1.2303563344307904}, {\"i\": 4, \"param\": \"vote_ideologies\", \"value\": 1.4095699020236472}, {\"i\": 5, \"param\": \"vote_ideologies\", \"value\": 1.5026541924075538}, {\"i\": 6, \"param\": \"vote_ideologies\", \"value\": 1.683459449630872}, {\"i\": 7, \"param\": \"vote_ideologies\", \"value\": 1.8074855216893626}, {\"i\": 8, \"param\": \"vote_ideologies\", \"value\": 2.163009422745624}, {\"i\": 9, \"param\": \"vote_ideologies\", \"value\": 2.3878209245231545}, {\"i\": 10, \"param\": \"vote_ideologies\", \"value\": 2.598320104840746}, {\"i\": 11, \"param\": \"vote_ideologies\", \"value\": 2.688778376408092}, {\"i\": 12, \"param\": \"vote_ideologies\", \"value\": 2.8661867112300907}, {\"i\": 13, \"param\": \"vote_ideologies\", \"value\": 3.201664318645701}, {\"i\": 14, \"param\": \"vote_ideologies\", \"value\": 2.7214855735308303}, {\"i\": 15, \"param\": \"vote_ideologies\", \"value\": 2.4867021671789207}, {\"i\": 16, \"param\": \"vote_ideologies\", \"value\": 2.472875250210437}, {\"i\": 17, \"param\": \"vote_ideologies\", \"value\": 2.1421931833532732}, {\"i\": 18, \"param\": \"vote_ideologies\", \"value\": 1.9487982145636569}, {\"i\": 19, \"param\": \"vote_ideologies\", \"value\": 1.7244384678806526}, {\"i\": 20, \"param\": \"vote_ideologies\", \"value\": 1.5343845533619274}, {\"i\": 21, \"param\": \"vote_ideologies\", \"value\": 1.3494125069761913}, {\"i\": 22, \"param\": \"vote_ideologies\", \"value\": 1.265922669690926}, {\"i\": 23, \"param\": \"vote_ideologies\", \"value\": 1.1029453747165046}, {\"i\": 24, \"param\": \"vote_ideologies\", \"value\": 1.0137779792881274}, {\"i\": 25, \"param\": \"vote_ideologies\", \"value\": 0.9027392141987254}, {\"i\": 26, \"param\": \"vote_ideologies\", \"value\": 0.7808764912438394}, {\"i\": 27, \"param\": \"vote_ideologies\", \"value\": 0.7168259100751055}, {\"i\": 28, \"param\": \"vote_ideologies\", \"value\": 0.45555181931154215}, {\"i\": 0, \"param\": \"legislator_ideologies\", \"value\": 3.0274116625749024}, {\"i\": 1, \"param\": \"legislator_ideologies\", \"value\": 2.766279759874206}, {\"i\": 2, \"param\": \"legislator_ideologies\", \"value\": 2.4653438814065205}, {\"i\": 3, \"param\": \"legislator_ideologies\", \"value\": 2.251108602962943}, {\"i\": 4, \"param\": \"legislator_ideologies\", \"value\": 1.9828997167202362}, {\"i\": 5, \"param\": \"legislator_ideologies\", \"value\": 1.6871869660807393}, {\"i\": 6, \"param\": \"legislator_ideologies\", \"value\": 1.442601280048146}, {\"i\": 7, \"param\": \"legislator_ideologies\", \"value\": 1.1961273914586594}, {\"i\": 8, \"param\": \"legislator_ideologies\", \"value\": 0.9840046698774239}, {\"i\": 9, \"param\": \"legislator_ideologies\", \"value\": 0.7702887325434579}, {\"i\": 10, \"param\": \"legislator_ideologies\", \"value\": 0.5797440529499934}, {\"i\": 11, \"param\": \"legislator_ideologies\", \"value\": 0.400551163901209}, {\"i\": 12, \"param\": \"legislator_ideologies\", \"value\": 0.23631347427358643}, {\"i\": 13, \"param\": \"legislator_ideologies\", \"value\": 0.07345155788236508}, {\"i\": 14, \"param\": \"legislator_ideologies\", \"value\": -0.10344339838385486}, {\"i\": 15, \"param\": \"legislator_ideologies\", \"value\": -0.26161359472066487}, {\"i\": 16, \"param\": \"legislator_ideologies\", \"value\": -0.42187707030420724}, {\"i\": 17, \"param\": \"legislator_ideologies\", \"value\": -0.5944109221973217}, {\"i\": 18, \"param\": \"legislator_ideologies\", \"value\": -0.7792006363012691}, {\"i\": 19, \"param\": \"legislator_ideologies\", \"value\": -1.0293319543607375}, {\"i\": 20, \"param\": \"legislator_ideologies\", \"value\": -1.296296864587514}, {\"i\": 21, \"param\": \"legislator_ideologies\", \"value\": -1.5690512975572333}, {\"i\": 22, \"param\": \"legislator_ideologies\", \"value\": -1.7010125361888917}, {\"i\": 23, \"param\": \"legislator_ideologies\", \"value\": -1.9430062038337634}, {\"i\": 24, \"param\": \"legislator_ideologies\", \"value\": -2.0558486367399333}, {\"i\": 25, \"param\": \"legislator_ideologies\", \"value\": -2.2292882589024132}, {\"i\": 26, \"param\": \"legislator_ideologies\", \"value\": -2.371797793226603}, {\"i\": 27, \"param\": \"legislator_ideologies\", \"value\": -2.482480281500808}, {\"i\": 28, \"param\": \"legislator_ideologies\", \"value\": -2.650137335418494}, {\"i\": 29, \"param\": \"legislator_ideologies\", \"value\": -2.760793952174754}, {\"i\": 0, \"param\": \"vote_biases\", \"value\": -3.1915079237657333}, {\"i\": 1, \"param\": \"vote_biases\", \"value\": -2.976076188960744}, {\"i\": 2, \"param\": \"vote_biases\", \"value\": -2.53637144038333}, {\"i\": 3, \"param\": \"vote_biases\", \"value\": -2.4091718258688477}, {\"i\": 4, \"param\": \"vote_biases\", \"value\": -2.2797188183618124}, {\"i\": 5, \"param\": \"vote_biases\", \"value\": -2.001203450965056}, {\"i\": 6, \"param\": \"vote_biases\", \"value\": -1.8477361302448365}, {\"i\": 7, \"param\": \"vote_biases\", \"value\": -1.597329482666332}, {\"i\": 8, \"param\": \"vote_biases\", \"value\": -1.5238047752970125}, {\"i\": 9, \"param\": \"vote_biases\", \"value\": -1.266992718187132}, {\"i\": 10, \"param\": \"vote_biases\", \"value\": -0.9690709511212404}, {\"i\": 11, \"param\": \"vote_biases\", \"value\": -0.621657182845019}, {\"i\": 12, \"param\": \"vote_biases\", \"value\": -0.26789680391863296}, {\"i\": 13, \"param\": \"vote_biases\", \"value\": 0.15043533131246672}, {\"i\": 14, \"param\": \"vote_biases\", \"value\": 0.44562032137378466}, {\"i\": 15, \"param\": \"vote_biases\", \"value\": 0.7214035508857959}, {\"i\": 16, \"param\": \"vote_biases\", \"value\": 1.105209089166033}, {\"i\": 17, \"param\": \"vote_biases\", \"value\": 1.2229450558809474}, {\"i\": 18, \"param\": \"vote_biases\", \"value\": 1.3992488585394502}, {\"i\": 19, \"param\": \"vote_biases\", \"value\": 1.4900281541134788}, {\"i\": 20, \"param\": \"vote_biases\", \"value\": 1.5784463829540871}, {\"i\": 21, \"param\": \"vote_biases\", \"value\": 1.6373244528243356}, {\"i\": 22, \"param\": \"vote_biases\", \"value\": 1.8330916441135159}, {\"i\": 23, \"param\": \"vote_biases\", \"value\": 1.8728115968186934}, {\"i\": 24, \"param\": \"vote_biases\", \"value\": 2.0530721706544233}, {\"i\": 25, \"param\": \"vote_biases\", \"value\": 2.206767630626554}, {\"i\": 26, \"param\": \"vote_biases\", \"value\": 2.3757598160965525}, {\"i\": 27, \"param\": \"vote_biases\", \"value\": 2.7741518265435325}, {\"i\": 28, \"param\": \"vote_biases\", \"value\": 3.0489465179870168}]}};\n",
       "var selector = \"#c113c9cd-ea77-4159-9de3-17dfd10cf854\";\n",
       "var type = \"vega-lite\";\n",
       "\n",
       "var output_area = this;\n",
       "require(['nbextensions/jupyter-vega/index'], function(vega) {\n",
       "  vega.render(selector, spec, type, output_area);\n",
       "}, function (err) {\n",
       "  if (err.requireType !== 'scripterror') {\n",
       "    throw(err);\n",
       "  }\n",
       "});\n"
      ]
     },
     "metadata": {
      "jupyter-vega": "#c113c9cd-ea77-4159-9de3-17dfd10cf854"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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"
     },
     "metadata": {
      "jupyter-vega": "#c113c9cd-ea77-4159-9de3-17dfd10cf854"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [22] used 0.000 MiB RAM in 0.019s, peaked 0.000 MiB above current, total RAM usage 4214.062 MiB\n"
     ]
    }
   ],
   "source": [
    "plot_params(g.params)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It will keep improving the log likelihood, but it will keep this basic pattern... Which is wrong! Like I said above for PyMC3 ADVI. "
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [default]",
   "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": "3.5.2"
  },
  "notify_time": "10",
  "toc": {
   "nav_menu": {
    "height": "146px",
    "width": "271px"
   },
   "navigate_menu": true,
   "number_sections": false,
   "sideBar": true,
   "threshold": 4,
   "toc_cell": false,
   "toc_section_display": "block",
   "toc_window_display": false
  },
  "widgets": {
   "state": {
    "d236125aff9d4976a5ba7f32f964924f": {
     "views": [
      {
       "cell_index": 23
      }
     ]
    }
   },
   "version": "1.2.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}