umap_with_imdb.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load a  previously trained doc2vec model\n",
    "\n",
    "The model was trained using gensim with the following parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import umap\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "from gensim.models import Doc2Vec\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# You will first need to unzip the file\n",
    "model = Doc2Vec.load('imdb.d2v')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## As a test case we are only going to take the training set of positive and negative documents"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "positive_examples = []\n",
    "negative_examples = []\n",
    "for i, x in enumerate(model.docvecs.doctags):\n",
    "    if x.startswith('TRAIN_POS'):\n",
    "        positive_examples.append(model.docvecs[i])\n",
    "    if x.startswith('TRAIN_NEG'):\n",
    "        negative_examples.append(model.docvecs[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "pos = pd.DataFrame(positive_examples)\n",
    "pos.insert(loc=0, column='sentiment', value=1)\n",
    "\n",
    "neg = pd.DataFrame(negative_examples)\n",
    "neg.insert(loc=0, column='sentiment', value=0)\n",
    "reviews = pd.concat([pos, neg], ignore_index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Let's try to embed the documents which are embedded into a 100 dimensional space into a two dimensional space"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 6min 13s, sys: 988 ms, total: 6min 14s\n",
      "Wall time: 4min 44s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "review_array = np.array(reviews.iloc[:, 1:])\n",
    "review_embedding = umap.UMAP().fit_transform(review_array)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df = pd.DataFrame(review_embedding, columns=['x', 'y'])\n",
    "df['labels'] = reviews['sentiment']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7f11ee803550>"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAD3CAYAAADi8sSvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkhJiqEtSQgx1SUqIoS5JCfn/K5zyVU0Ycq4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f11fe79ff60>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(x=df['x'], y=df['y'], c=df['labels'], cmap='tab10')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This doesn't look great. Let's try decreasing the size of the local neighborhood and increasing min_dist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 4min 24s, sys: 364 ms, total: 4min 24s\n",
      "Wall time: 3min 28s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "tweaked_embedding = umap.UMAP(n_neighbors=10, min_dist=1.0).fit_transform(review_array)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df2 = pd.DataFrame(tweaked_embedding, columns=['x', 'y'])\n",
    "df2['labels'] = reviews['sentiment']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7f65d348b240>"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAD3CAYAAAAT+Z8iAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAH49JREFUeJzt3XmcXFWd9/HPrbW36i1d6SQdsodDgABRsoAIvBwQRBxw\nRB5evNzRlzyPo4g64+jIKA6O+4LPqJlx4jjzgA+OIqMPijADssgStrBGTkggkKQ76e70Ur1W13Kf\nP6qSdKer00mnku469X3/1XWq+t7zS3e+9/a5597j+b6PiIi4KzDdHRARkWNLQS8i4jgFvYiI4xT0\nIiKOU9CLiDguNN0dKKSjo2/MVKCGhiq6uwenqzvHhGs1uVYPqKZS4VpNR1NPPB7zCrWXxBl9KBSc\n7i4UnWs1uVYPqKZS4VpNx6Kekgh6ERGZOgW9iIjjFPQiIo5T0IuIOG5GzrqZiYZSGdr7ksRrolRF\nchdLtg4nebh/kNmhEG+tqyHoFbzgLSIyrRT0QNb3Sfk+Ec/DOyisM1mf7z2wjQe27qUtkaS5Jso5\nSxtJnBjjvv5BBrK5maD/2tnDF+Y2cVp15f7v9X2fdNYnHJzaH06+7/No/xCvjoywprqS5RXRqRcp\nImWrrIO+N53mo9vb2JocIeNDYyjI/2yq54qm+v2f+f4Dr3Db0637X+/pT3L7s22kuxKkT67H6xwm\nuGuQrckMn6hq55tvWsKqeXX87wdf5ZHtXSSG05xQX8nlK+dw6alzDrtvO5MpbtjVzqbBYTJAdcDj\n3Fg1fz8vzu50Gt+HhdHwuAOTiMjByjbo077PZS/vYG8mu7+tPZ3hy7v3UhkM8vaGGMl0lv/a2lnw\n+4M7BgjsHMDzYV/UDnSP8Olfb+akpio27erb/9muwRQvtPfRmclw1clz2LSzl+XpLE2hic/0v7Sz\nnU3PtxPoShLI+iRjYe5alOLhze0M7xzAS2aoqgrz9hWz+ezaRcX4JxERR5Vt0K9v7xoT8vv4wA87\nunh7Q4w/dPbS0Zcs+P0e4BV4lP9QMsOzrX3j2jNpnx89vJ1fPL6T9kSSSMjj9Ll1vHdNC7OqIixt\nqiEY8EhlfX7f08emR3cR3jN8YAO9KQLtwyRHsvuvoA8NZfjlw69TnYW/PGvREf8biEh5KNugf6R/\n4luMO1NpAO4cHCRbESQwlDmibWcnWMslM5ShPb+tkbTPEzt6eGJHDwANDVFWnjiLLU0hXn+xc2zI\n5wVGxh+Y8OGOF3dz7ZoFhI7gWsB/diW4JzFAdybN/HCEq2bV8sZR1xdExB1lG/SzgiFgpOB7FYFc\nYG5Npck0VxLY3l+UfR5qNL27O8kDG1vxgfAknz1YIjHC7r4k8+sPL6h/1N7Fj9u7SeVfvzA0wuMD\nQ9w0P86bY9VHsGcRKQVlO4/+mnj9hMWfmw+7qoBH5sRa0rMr8L3csM6x5JH7gRzp5dVwJEB9Zfiw\nPjuQyXJHd2J/yO/Tlclwy97eI9yziJSCsg36M6oreV9jHQc/PmhlRZQvt8QBWF1dCYNpAj0jYy66\nTpeJDjSxeBU10cP74+zh/kHaUoWHol4eHiGjNYRFnFO2QzcAn57XxFVNdfxbRw+9mQzvbarn1KqK\n/e9/cs4sfv3oDjKFxsanyb4Y9vJfZ5uiZFbU4/v+YU21jIeCBIFCUV8V8Mr3yC/isLIOeoCWSJjP\n58/gD1bheWS6Cs+6mQ6FYtyPBgiGIen7hH14YGsn7f1Jzl0yi3kFxuzPqKrgtMoom4bG17W6ulLz\n8kUcVPZBfyh/TAySHUjN2LNcDwjuGqJt9iBv6d5C5KVeBrpzAf7jR17jwpPi/PWfLScwKrw9z+Nv\n5jXxdzs7sMncxegQsLamkr+a0zQNVYjIsaagP8iTuxPc9twuvKTP8539BI5sZuVx5wHBPUMk+9Kk\n+g5cYk0kM9z+7G7m1lby/jUnjPmeFRVR1rc0888v7SZbFeT8lnreVFOls3kRR5Vt0D8zMMTP9ibY\nPjJCbTDA2dVVPLhtLy8+uRtm0Jj84fD6RvD6Ch+Rfrulg/etnk/Gh/V/3M4fX+mkNZEknfFJZX0i\nQY/d8+tYfMFyWg5zeqaIlJayDPonB4b47I49tKcPhOPG/iHCL3QQLLGQB6A/M+GMoG0Dw1y5bSep\n5/ay69XEuPdHMj4bX+vhy3dvYf2Vp+msXsRBM3X4+Zi6pbNnTMgDeH0pAomDZ5eXhqA/8dRLvzrE\nS12D7Nwx/rEMoz3X2svjr/UUv3MiMu3KMuhfTo6/I9bbO3Nm10xFofPwbCRAekE1wa4kXvrQ8+PT\nWXi2VTdMibioLIO+OjC2bK87SXhbYtpviComH0gtj+E3VpCtDuEfRnG3PLGTT//nC7T3jX/OjoiU\nrrIM+nU1+YuOvk/g1QThJzrxZvjsmiPlAeFXc8/o8RuiZBsmX7RkKJ3lwW1d3Pj7Lfi6Q1bEGWUZ\n9B+fPYvmNEQe7yC8pY+Ao5nmDWb2D0mlTq0n0xjdf2Yf8CZ+pMOmnT08+brG60VcUZZBHw54LN42\nQKAn5dRwzcE8INCTv/ZQGSK1uomRNU2Y1c385sNrOGGC6ZSpLGztHDh+HRWRY8qp6ZUj2dzar9WT\nPJe9dyjFq7uL8+jhmc5r68dbHGNBOMiKmkrWzovzzoYYQc9jTl2U13uGxn1PRSjAynm109BbkfLk\n+z6Pv9aDbe/nzGVNrGioKOpUZyeCvjud5htte3lqYIihrM/yigjvmVXHW+pqCn5+MJVhcIInOLom\nOOATvq+N4booLSfP5p1rmvnvxADtqTRnndjE860JhlJj7x1YvbCeU+cq6EWOh66BEW743Us8vbOH\ndBbCD2/njJZabnr7ChqrI0XZx3EJemNMAPghcDqQBD5srd1ajG1nfZ9P79jDEwMHZoo8MTjMy8kR\nqoMB1tZUjfueWDREXUWYzoHCC4+4xAP8jE9v1zC3Pvw6d3Qn6Jqfe95+bSiAOT1OescAr3cPEYuG\nWL2gns+8Zdn0dlqkjHzjvq08PuqaWCrr88SOXr5+71a+/ucnF2Ufx+uM/nKgwlp7ljFmHfBt4LJi\nbPjexABPDYyfDtiTyfLLrr4xQZ/2fXb3Jfnb32wui5Afx4e+HQPQUgWeRyKb5YlZIT6xYhFX1NRQ\nFQkRPcSC5SJSXInhFE9NMPHh6R099A6lqDvMRYUO5XgF/TnA7wGstY8ZY8481IcbGqoIhcYuCRKP\nxwp+dmdfPxM9tKCTLPF4jMFMli++vIsHe/rYu6mD9J7yGJ8vJDCUhowPoQPjfw8ODvP5UxYe9bYn\n+hmVMtVUGkq1puGuQfpHCg8j9yXThKujxGcd/fKexyvoa4HRt11mjDEha2260Ie7u8cu3B2Px+jo\nKHwLf11q4mfTxHzo6OjjutfauK8vt81wd3LcqlLlxI8EITj2Io8dGJ7w3/dwHepnVKpUU2ko5ZrC\nWZ/Fsyp5uWNw3HuLZ1URSWeOqLaJDnjH6+/0BDC6B4GJQv5IvaMhxoqK8Rcsoh5cUh/jhcFhHukf\n/49YrjJNFXDQ1fxh32fTwPjZNyJybAUDHu9cOZeKg4ZMoyGPy1fOJRQozsyb43VG/zDwDuA/8mP0\nzxdrw2HP4ysts/na7r08MzDECLAoEuLdjXVcXFfD/+nsYXjUDVHZ+gjBEn142dHwgcycCjInjp9N\n45NbL3ZVtR5TLHK8vXtVC7GKEL/b3E57f5L5jdX82dJG3nZyc9H2cbyC/g7gQmPMI+QmgnywmBtf\nXhllw+J5bBsaoTuT4bSqCiL5I+GSaGTMGqmZpTECiRTBngMXY32mf+HvY80D/Opw7pbYg9QGPNYV\nmJ0kIsfHxSuauXhFLtiPxVDUcQl6a20WuPZY72dp5fghnLNrKllVVcGTg/mZOZEgqdVNZHf04yVS\n+EEItg1P+nRHJ2QL13h+rJoF0aO/si8iM5Pzc+k8z+Omltm8qaaSivzJbDTokVkYI31KA17SL4uQ\n9wOQbaoAoDEYIB4Ksiwa5v2z6vjS/NnT3DsROZacuDN2Mi3RMOsXzePV5AhtqRQjGZ8vtnbS+1IX\noXb3H8nrA9G5VZy5oJ6zqit5b1M9PrmjvFaUEnFfWQT9PoujERZHc8M7N/nwmftbJ5yD7wrfg/Ti\nGoIn1vHReIMuuIqUobIJ+qFsll91JehKZzi9qoLFXoBssigzPGc0zwd8GPThNz19CnqRMlQWQb+x\nf5CbWjvZPpKbVhkETDhMtiJEYMD9sCeTuwYxkHX97xcRKcT5i7EZ3+dbu/fuD3nITbXcnEqRaa6c\ncFFtV/hAtiFCoHWQrhe7uP3ZVkbSCnyRcuL8Gf39iQFeGi78ALPMstzNusE9Q3jDafygB2kfL+vO\nvPrMrAih7f0Ee1M8Czz7Yie3P9vGFy8+ETO7NJ8PIiJHxvkz+u7MIZ4773lkltcycvZsUsvr8LIQ\ncCjkI/URaiIhgr1j7wR+uWOA793/yjT1SkSON+eD/oJYNU2TrDiFB8HWwZKbT+9PUtaicJjavsLX\nIJ5tTfDqXi0XKFIOnA/6+nCIv2isHTdG1RwKcFplNDeXfDBDoMSef+N7kDyziXTjxCvQBDyP4QnG\n41MZn97hMrgQLSLuj9EDfLx5FidEwtzbO0Aim2VBJMTVs+o5qSLCfycGeGxvH78LtZMucEYf8CAS\nCjB80OOQp/v5OJ6f2//JZ8+j54E22vuS4z6zan4t1R1BntrRO+69xY2VnDpHY/Qi5aAsgh7g8oZa\nLm8Y/+TGC+tquLCuho75nTy6vXvc++csmcV7z5zPjX94mZ3tuccdZ6tDpOdV4g1nCO0YPGaBnyUX\n5oW2HwwF+OD8Rj66cDb3jIT5wUOv0jN04Az91DkxLjwxzhtOqOOVzgG6R71XGQ7wrjPmEZpsSEtE\nnFA2QT+ZT56/hK7fvYRtPzBuvaK5huvOW8KChkp+9Z43ctfrXdzTlSDYUEEsFOD2nn4CncMEh47N\ndEUv5JGZXUGodfyz4s9b1Mj1i+cAcPnKubyhpY5fP7+b7V2DbO8a4E97+vjwbc9yUnMN717Vwo7u\nIdoSwzRWRXjbitmcv7zpmPRZRGaesg963/d5Zlcvnf0j/OO7VnLvlk5aE8PMr6/kHac07z/r9TyP\nSxbO4pKFswCwQ0nu6OnHr43A0DF6Xk7WJ9sYJe1DuGMYP+1TGQrwxgX1fO6tYxfwXtBYxfvWzOcD\nP3uGnT0HhnE27+mnLTHM99+1kpOaNVQjUo7KOuj/tLuPb/1hGy+2Jcj40ByL8o5Tmvn4uUsm/d7l\nFRFWVETY3FJFsH0496iBIvOyELa9NK2ezRfespy9HUOY2TUQi7DX86nz/TEPJfuPZ9rY2TP+oNM9\nlOaO59r43IUKepFyVLaDtOlMlpvu2cJzrYl9TwhgT1+Snz7+Or9+vm3S7w94HtfEG2icU0O28tgd\nL72Uzxu6s5wzp56KE2J8sa+HK1/ZyRVbd/L+V3bxxKhlEjv7C98YBtA5MPF7IuK2sg36323ew5aO\n8fPI01m4b0vnYW3jwroavtncRGSCVdw9cms/7pMNe4UfuRCAcM3EB4vegRE2Dw7zlbYONg+P4ANp\nYNNQkr/b1UFPKnehdV5tdMJtNMcmfk9E3Fa2Qd9+iLPf7qHDn1O/tCJCRaDwP2Mo6LH+ytP5wRUr\nueCypYy8ZR6pU+rJVh8I9UxNiHVntXDd6oUT7qOpJsrt3X10Z8Zf9N2ZSvOz7gSQW3tyyazxSwI2\n10S4clXLYdckIm4p26A/dW6M0ATVz6+vOOztNFZFOGWC+einzIlxypwYaxY2kIoEAcjOr2bk7NmM\nnN5A8oxGUmfNpmF+jMtPm8uypupx26itCHHZqXPoTE98c1NH/oy+KhLkHy5dwZuXNFJfEaImEmT1\ngnpuuNiwqFFrwoqUq7K9GLt2YQNrFjbyyKtdY9obq8Jccfq8I9rWR85eyM6eIVoTB2a7zIlF+fBZ\nC/dfLB2zJmvAIzvnQPAuioaJhgL8/SUn8f0HX+GZXb0k01lOml3DVW9s4Q0n1HN368R/ZbSED2x7\naVM133nnqfQn02SyPnWVWgtWpNx5vj/znu/S0dE3plPHYlV0gKFUhu/d/wpP7ehhcCTDsng1V72h\nhbMXNx7xtjr6k/z86V2094/QVB3mylXzmTNqzLw7leZD21vZmhwb2CdXRPjpkhYqRw3/dPYnGU5n\nmVdXQSB/oHgtmeIj23fRlhp7PWBZNMKtS1qomuabn47Vz2g6qabS4FpNR1NPPB4reP9m2Z7RA1SG\ng3zuwuX4vp9bQ/Uo1k+N10T5y0NMy2wIh/jugjmsb+/mhaFhQsEAp0QjfLy5cUzIQ25M/mALo2Fu\napnNhs4eXhxKEgROq6rguubGaQ95EZnZyjro9/E877g8t2ZRNMLXTmgGpnbUXlNTxZqaKhKZDEE8\nqhXwInIYFPQlqDYYnO4uiEgJ0SmhiIjjFPQiIo5T0IuIOK5oY/TGmDrgFqAWiACfstY+aoxZB9xM\n7q79e6y1NxZrnyIiMrlintF/CrjXWnse8AHgB/n29cDVwDnAWmPMqiLuU0REJlHMWTffBfbdGhoC\nho0xtUDUWrsNwBhzN3ABsKmI+xURkUOYUtAbY64Brj+o+YPW2ieMMXPIDeF8ktwwTmLUZ/qASR/2\n3tBQRSg0dgphPO7es9Rdq8m1ekA1lQrXaip2PVMKemvtBmDDwe3GmJXAbcBnrLUP5M/oR/c4BvRM\ntv3u7sExr127xRncq8m1ekA1lQrXajrKRyAUbC/aGL0x5mTgF8DV1tq7AKy1CWDEGLPUGOMBFwEP\nFWufIiIyuWKO0X8VqABuNsYA9FprLwOuBW4FguRm3Wws4j5FRGQSRQv6fKgXan8MWFes/YiIyJHR\nDVMiIo5T0IuIOE5BLyLiOAW9iIjjFPQiIo5T0IuIOE5BLyLiOAW9iIjjFPQiIo5T0IuIOE5BLyLi\nOAW9iIjjFPQiIo5T0IuIOE5BLyLiOAW9iIjjFPQiIo5T0IuIOE5BLyLiOAW9iIjjFPQiIo5T0IuI\nOE5BLyLiOAW9iIjjFPQiIo5T0IuIOE5BLyLiOAW9iIjjQsXeoDHmJGAj0GytHTbGrANuBtLAPdba\nG4u9TxERmVhRz+iNMbXAt4HkqOb1wNXAOcBaY8yqYu5TREQOrWhBb4zxgH8GPg8M5ttqgai1dpu1\n1gfuBi4o1j5FRGRyUxq6McZcA1x/UPNrwG3W2meNMfvaaoHEqM/0AUsm235DQxWhUHBMWzwem0pX\nZzTXanKtHlBNpcK1mopdz5SC3lq7Adgwus0YsxW4Jn8QmAPcA1wKjO5xDOiZbPvd3YNjXsfjMTo6\n+qbS1RnLtZpcqwdUU6lwraajqWeiA0TRLsZaa5ft+9oYsx14a/5i7IgxZinwCnARoIuxIiLHUdFn\n3RRwLXArECQ362bjcdiniIjkHZOgt9YuGvX1Y8C6Y7EfERGZnG6YEhFxnIJeRMRxCnoREccp6EVE\nHKegFxFxnIJeRMRxCnoREccp6EVEHKegFxFxnIJeRMRxCnoREccp6EVEHKegFxFxnIJeRMRxCnoR\nEccp6EVEHKegFxFxnIJeRMRxCnoREccp6EVEHKegFxFxnIJeRMRxCnoREccp6EVEHKegFxFxnIJe\nRMRxCnoREccp6EVEHBcq1oaMMUHgO8CZQBT4krX2TmPMOuBmIA3cY629sVj7FBGRyRXzjP69QNha\n+ybgMmBZvn09cDVwDrDWGLOqiPsUEZFJFO2MHrgIeMEY81vAAz5ujKkFotbabQDGmLuBC4BNRdyv\niIgcwpSC3hhzDXD9Qc0dwDBwKXAu8K/kzuQToz7TByyZbPsNDVWEQsExbfF4bCpdndFcq8m1ekA1\nlQrXaip2PVMKemvtBmDD6DZjzG3AndZaH3jAGHMiuZAf3eMY0DPZ9ru7B8e8jsdjdHT0TaWrM5Zr\nNblWD6imUuFaTUdTz0QHiGKO0f8RuATAGHM68Lq1NgGMGGOWGmM8csM7DxVxnyIiMolijtH/GPiR\nMeYxcmP01+bbrwVuBYLkZt1sLOI+RURkEkULemttEvhQgfbHgHXF2o+IiBwZ3TAlIuI4Bb2IiOMU\n9CIijlPQi4g4TkEvIuI4Bb2IiOMU9CIijlPQi4g4TkEvIuI4Bb2IiOMU9CIijlPQi4g4TkEvIuI4\nBb2IiOMU9CIijlPQi4g4TkEvIuI4Bb2IiOMU9CIijlPQi4g4TkEvIuI4Bb2IiOMU9CIijlPQi4g4\nTkEvIuI4Bb2IiOMU9CIijlPQi4g4LlSsDRlj6oDbgBogCbzHWrvbGLMOuBlIA/dYa28s1j5FRGRy\nxTyj/wDwvLX2zcDPgb/Kt68HrgbOAdYaY1YVcZ8iIjKJYgb980As/3UtkDLG1AJRa+02a60P3A1c\nUMR9iojIJKY0dGOMuQa4/qDmjwFvNcZsBhqBN5ML/MSoz/QBSybbfkNDFaFQcExbPB6b4NOly7Wa\nXKsHVFOpcK2mYtczpaC31m4ANoxuM8b8CviGtfafjDGnAbeTG64Z3eMY0DPZ9ru7B8e8jsdjdHT0\nTaWrM5ZrNblWD6imUuFaTUdTz0QHiGIO3XQDvfmv24Faa20CGDHGLDXGeMBFwENF3KeIiEyiaLNu\ngBuAfzHG/C8gDHwk334tcCsQJDfrZmMR9ykiIpMoWtBba1uBSwq0PwasK9Z+RETkyOiGKRERxyno\nRUQcp6AXEXGcgl5ExHEKehERxynoRUQcp6AXEXGcgl5ExHEKehERxynoRUQcp6AXEXGcgl5ExHEK\nehERxynoRUQcp6AXEXGcgl5ExHEKehERxynoRUQcp6AXEXGcgl5ExHEKehERxynoRUQcp6AXEXGc\ngl5ExHGh6e5AOdnS3s8vnmlldyLJvMYqLlzWyJkLGqa7WyLiOAX9cbLxtW6+/HtLe/9IruG1bv5r\n8x6uO28xl62cO72dExGnaejmOPn3x3ccCPm8vmSa//v0LtJZf5p6JSLlQEF/HAyOZNjS3l/wvW2d\ng7zQljjOPRKRcnJUQzfGmHcC77bWXp1/vQ64GUgD91hrb8y3fxF4e779k9bax4+q1yUmGPAIhwof\nU0MBqAoHj3OPRKScTDnojTE3AxcBz4xqXg+8C3gF+K0xZhXgAecBa4ETgNuB1VPdbymKhgKcNq+W\ne7d0jnvvlLm1LI9XT0OvRKRceL4/tfFhY8z/ANqBj1prrzLG1AIbrbUr8u9fB0SAJFBlrf1avn0T\n8FZrbcdE206nM34o5NZZ7o6uQT5269M8t6t3f9vipiq+ecXpnLmocRp7JiIO8Qo1TnpGb4y5Brj+\noOYPWmt/bow5f1RbLTB6sLkPWAIMA3sPaq8DJgz67u7BMa/j8RgdHX2TdXVGqwDWv3slv36+jde6\nh1nUXMPFy2ZRHQmVfG3gxs/oYKqpNLhW09HUE4/HCrZPGvTW2g3AhsPYRwIYvZcY0AOMTNBedsLB\nAFec0QK498spIjNX0WbdWGsTwIgxZqkxxiM3fv8Q8DBwkTEmYIxZAASsteMHq0VE5Jgo9g1T1wK3\nAkFys242AhhjHgIeJXdg+ViR9ykiIodwVEFvrb0fuH/U68eAdQU+9yXgS0ezLxERmRrdMCUi4jgF\nvYiI46Y8j15EREqDzuhFRBynoBcRcZyCXkTEcQp6ERHHKehFRBynoBcRcZyCXkTEcTN2cXBjTB1w\nC7nHH0eAT1lrH51oFatSYIwJAD8ETif3nP4PW2u3Tm+vjpwxJgz8BFgERIGbgM3ATwEfeAH4mLU2\nO01dnDJjzGzgKeBCcr9jP6WEazLGfA74c3L/h34IPECJ1pT/vfs3cr93GeAjlPDPyBizFvi6tfZ8\nY8wyCtRhjPkI8FFydd5krb1zKvuayWf0nwLutdaeB3wA+EG+fT1wNXAOsDa/ilWpuByosNaeBfwN\n8O1p7s9UvQfYa619M3Ax8I/Ad4Av5Ns84LJp7N+U5IPkn4ChfFNJ15RfL+Js4E3kVnk7gdKu6RIg\nZK09G/gy8BVKtB5jzF8D/0JuqQooUIcxZg7wCXI/v4uArxpjolPZ30wO+u+S+08Hub88hvOrWEWt\ntdustT5wN3DBdHVwCs4Bfg/7HwB35vR2Z8p+AdyQ/9ojd7bxRnJniwB3UVo/l32+Re5EojX/utRr\nugh4HrgD+H/AnZR2TVuAUP4v41ogRenWsw34i1GvC9WxBnjYWpu01vYCW4HTprKzGTF0c4hVrJ7I\nH9VuAT7JxKtYlYpaoHfU64wxJmStTU9Xh6bCWtsPYIyJAb8EvgB8K3/whQOriJUMY8wHgA5r7d35\n4Q4Ar5RrApqAhcClwGLgN+TWgyjVmvrJDdu8RK62S4FzS7Eea+3txphFo5oK/a4dnBdTrm9GBP1E\nq1gZY1YCtwGfsdY+kD+jL+XVqg5ehStQaiG/jzHmBHJnij+01v7MGPONUW+X2s8F4EOAb4y5ADgD\n+Hdg9qj3S7GmvcBL1toRwBpjhskN3+xTajVdD9xtrf1c/vfvPnLXHvYptXpGG31dYV8dE63ad8Rm\n7NCNMeZkckMEV1tr74JDrmJVKh4mN85I/qLy89PbnakxxjQD9wCftdb+JN+8adQawm+jtH4uWGvP\ntdaeZ609H3gGeB9wVynXBPwRuNgY4xlj5gHVwL0lXFM3B85wu4AwJf57N0qhOh4H3myMqchPTllB\n7kLtEZsRZ/QT+Cq5CxU3G2MAeq21lzHBKlYl4g7gQmPMI+TGtj84zf2Zqs8DDcANxph9Y/XXAd83\nxkSAP5Eb0il1nwZ+XKo1WWvvNMacSy4w9q3u9iqlW9N3gZ/kV6yLkPs9fJLSrWe0cb9r1tqMMeb7\n5EI/APyttXZ4KhvXY4pFRBw3Y4duRESkOBT0IiKOU9CLiDhOQS8i4jgFvYiI4xT0IiKOU9CLiDju\n/wM4Bi3amxFOlwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f65d3633fd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(x=df2['x'], y=df2['y'], \n",
    "            c=df2['labels'], cmap='tab10')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It still looks basically the same. If we zoom in (to ignore the outlying points) we see"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7f65d33c74e0>"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAD7CAYAAABkO19ZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f65d34a9550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.axis([-20, 20, -20, 20])\n",
    "plt.scatter(x=df2['x'], y=df2['y'], c=df2['labels'], cmap='tab10')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Let's try PCA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.decomposition import PCA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 264 ms, sys: 8 ms, total: 272 ms\n",
      "Wall time: 157 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "review_array = np.array(reviews.iloc[:, 1:])\n",
    "pca_embedding = PCA(n_components=2).fit_transform(review_array)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7fb87e28fcc0>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAW8AAAD5CAYAAADodLT+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fb8818ba400>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pca_df = pd.DataFrame(pca_embedding, columns=['x', 'y'])\n",
    "pca_df['labels'] = reviews['sentiment']\n",
    "\n",
    "plt.scatter(x=pca_df['x'], y=pca_df['y'], c=pca_df['labels'], cmap='tab10')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Let's try it without the spectral initialization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 3min 1s, sys: 352 ms, total: 3min 1s\n",
      "Wall time: 3min 1s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "review_array = np.array(reviews.iloc[:, 1:])\n",
    "review_embedding = umap.UMAP(init='random').fit_transform(review_array)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "random_embedding = review_embedding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7f11fbe8a978>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAW8AAAD7CAYAAAClvBX1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f11fc27ccf8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "random_df = pd.DataFrame(random_embedding, columns=['x', 'y'])\n",
    "random_df['labels'] = reviews['sentiment']\n",
    "\n",
    "plt.scatter(x=random_df['x'], y=random_df['y'], c=random_df['labels'], cmap='tab10')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/home/gclenden/anaconda3/lib/python3.6/site-packages/umap-0.0.1-py3.6-linux-x86_64.egg/umap/__init__.py'"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "umap.__file__"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Let's try looking into why this didn't work"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from scipy.sparse import find\n",
    "from umap.umap_utils import fuzzy_simplicial_set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "u = umap.UMAP()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "50\n",
      "3\n"
     ]
    }
   ],
   "source": [
    "print(u.n_neighbors)\n",
    "print(u.oversampling)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1min 24s, sys: 116 ms, total: 1min 24s\n",
      "Wall time: 1min 24s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "graph = fuzzy_simplicial_set(review_array, u.n_neighbors, u.oversampling)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "row_idx, col_idx, non_zero_val = find(graph)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[-182.76703723076869,\n",
       " -182.76703723076869,\n",
       " -155.61499637503186,\n",
       " -155.61499637503186,\n",
       " -132.17115218684077,\n",
       " -132.17115218684077,\n",
       " -121.72089185206474,\n",
       " -121.72089185206474,\n",
       " -120.62853971293048,\n",
       " -120.62853971293048,\n",
       " -119.42351710640956,\n",
       " -119.42351710640956,\n",
       " -118.06674479569836,\n",
       " -118.06674479569836,\n",
       " -115.82457962375778,\n",
       " -115.82457962375778,\n",
       " -114.40917546049887,\n",
       " -114.40917546049887,\n",
       " -112.74276101187854,\n",
       " -112.74276101187854,\n",
       " -111.45909513242631,\n",
       " -111.45909513242631,\n",
       " -110.80671431346163,\n",
       " -110.80671431346163,\n",
       " -109.84968350614497,\n",
       " -109.84968350614497,\n",
       " -109.63008147114049,\n",
       " -109.63008147114049,\n",
       " -109.5757808450868,\n",
       " -109.5757808450868,\n",
       " -109.4795346279987,\n",
       " -109.4795346279987,\n",
       " -109.11587072471065,\n",
       " -109.11587072471065,\n",
       " -107.25187171228663,\n",
       " -107.25187171228663,\n",
       " -105.54299282710988,\n",
       " -105.54299282710988,\n",
       " -105.53656502373077,\n",
       " -105.53656502373077,\n",
       " -104.81966846271192,\n",
       " -104.81966846271192,\n",
       " -104.51705211744323,\n",
       " -104.51705211744323,\n",
       " -101.96537283487224,\n",
       " -101.96537283487224,\n",
       " -100.70858111455843,\n",
       " -100.70858111455843,\n",
       " -100.05713477582779,\n",
       " -100.05713477582779,\n",
       " -99.757132353921378,\n",
       " -99.757132353921378,\n",
       " -97.982763355932207,\n",
       " -97.982763355932207,\n",
       " -97.419318458277061,\n",
       " -97.419318458277061,\n",
       " -97.333337136650186,\n",
       " -97.333337136650186,\n",
       " -97.26573797488372,\n",
       " -97.26573797488372,\n",
       " -97.177230380595262,\n",
       " -97.177230380595262,\n",
       " -96.433267756684558,\n",
       " -96.433267756684558,\n",
       " -96.146951339715045,\n",
       " -96.146951339715045,\n",
       " -95.529067142226793,\n",
       " -95.529067142226793,\n",
       " -95.097425950104395,\n",
       " -95.097425950104395,\n",
       " -94.703937600622297,\n",
       " -94.703937600622297,\n",
       " -93.295745028651368,\n",
       " -93.295745028651368,\n",
       " -92.924876247171795,\n",
       " -92.924876247171795,\n",
       " -92.605565345286507,\n",
       " -92.605565345286507,\n",
       " -92.351548095174863,\n",
       " -92.351548095174863,\n",
       " -91.899204644782358,\n",
       " -91.899204644782358,\n",
       " -91.811400408568375,\n",
       " -91.811400408568375,\n",
       " -91.778032529528886,\n",
       " -91.778032529528886,\n",
       " -91.724352427449048,\n",
       " -91.724352427449048,\n",
       " -91.61532529044193,\n",
       " -91.61532529044193,\n",
       " -91.417401193614552,\n",
       " -91.417401193614552,\n",
       " -91.285497205366411,\n",
       " -91.285497205366411,\n",
       " -90.2756574258038,\n",
       " -90.2756574258038,\n",
       " -89.75035901784625,\n",
       " -89.75035901784625,\n",
       " -89.391362706551647,\n",
       " -89.391362706551647,\n",
       " -89.157197237344022,\n",
       " -89.157197237344022,\n",
       " -89.12908907336525,\n",
       " -89.12908907336525,\n",
       " -89.067113413949116,\n",
       " -89.067113413949116,\n",
       " -88.666756990699511,\n",
       " -88.666756990699511,\n",
       " -87.327641793279028,\n",
       " -87.327641793279028,\n",
       " -87.223778405905009,\n",
       " -87.223778405905009,\n",
       " -87.184565401472639,\n",
       " -87.184565401472639,\n",
       " -86.210914380856451,\n",
       " -86.210914380856451,\n",
       " -86.124160093269879,\n",
       " -86.124160093269879,\n",
       " -85.724769267033935,\n",
       " -85.724769267033935,\n",
       " -85.041223358151711,\n",
       " -85.041223358151711,\n",
       " -84.960216770436872,\n",
       " -84.960216770436872,\n",
       " -84.3573134169703,\n",
       " -84.3573134169703,\n",
       " -84.032159795221162,\n",
       " -84.032159795221162,\n",
       " -83.859549500548766,\n",
       " -83.859549500548766,\n",
       " -83.69283266646238,\n",
       " -83.69283266646238,\n",
       " -82.740309425304375,\n",
       " -82.740309425304375,\n",
       " -82.691919173365534,\n",
       " -82.691919173365534,\n",
       " -82.291595821969338,\n",
       " -82.291595821969338,\n",
       " -82.148619279971186,\n",
       " -82.148619279971186,\n",
       " -82.044574959508452,\n",
       " -82.044574959508452,\n",
       " -81.822164074568107,\n",
       " -81.822164074568107,\n",
       " -81.539952594401456,\n",
       " -81.539952594401456,\n",
       " -81.039197700556002,\n",
       " -81.039197700556002,\n",
       " -80.804666327443442,\n",
       " -80.804666327443442,\n",
       " -80.236976678385815,\n",
       " -80.236976678385815,\n",
       " -80.025255103392695,\n",
       " -80.025255103392695,\n",
       " -79.763291797159738,\n",
       " -79.763291797159738,\n",
       " -79.572529502695616,\n",
       " -79.572529502695616,\n",
       " -79.409132023025208,\n",
       " -79.409132023025208,\n",
       " -79.048528962641129,\n",
       " -79.048528962641129,\n",
       " -78.923299903731831,\n",
       " -78.923299903731831,\n",
       " -78.778979264279414,\n",
       " -78.778979264279414,\n",
       " -78.591574519130432,\n",
       " -78.591574519130432,\n",
       " -78.091322568946168,\n",
       " -78.091322568946168,\n",
       " -77.937336000491172,\n",
       " -77.937336000491172,\n",
       " -77.897599201475188,\n",
       " -77.897599201475188,\n",
       " -77.542266226634752,\n",
       " -77.542266226634752,\n",
       " -77.139301397760036,\n",
       " -77.139301397760036,\n",
       " -77.100419359604061,\n",
       " -77.100419359604061,\n",
       " -76.83089204583608,\n",
       " -76.83089204583608,\n",
       " -76.621462821159511,\n",
       " -76.621462821159511,\n",
       " -76.292033302536879,\n",
       " -76.292033302536879,\n",
       " -76.271197177822685,\n",
       " -76.271197177822685,\n",
       " -75.788100298235193,\n",
       " -75.788100298235193,\n",
       " -75.648451928284004,\n",
       " -75.648451928284004,\n",
       " -75.460569025531697,\n",
       " -75.460569025531697,\n",
       " -75.085700615834256,\n",
       " -75.085700615834256,\n",
       " -74.870619343994179,\n",
       " -74.870619343994179,\n",
       " -74.82776238434613,\n",
       " -74.82776238434613,\n",
       " -74.732350660650937,\n",
       " -74.732350660650937,\n",
       " -74.560261901936727,\n",
       " -74.560261901936727,\n",
       " -74.547685054438489,\n",
       " -74.547685054438489,\n",
       " -74.501304468463843,\n",
       " -74.501304468463843,\n",
       " -74.302739571551683,\n",
       " -74.302739571551683,\n",
       " -74.274860546688458,\n",
       " -74.274860546688458,\n",
       " -74.26901012155696,\n",
       " -74.26901012155696,\n",
       " -74.250854530888006,\n",
       " -74.250854530888006,\n",
       " -73.963303275923465,\n",
       " -73.963303275923465,\n",
       " -73.82436048970095,\n",
       " -73.82436048970095,\n",
       " -73.782673315529152,\n",
       " -73.782673315529152,\n",
       " -73.724474624636812,\n",
       " -73.724474624636812,\n",
       " -73.365081013971732,\n",
       " -73.365081013971732,\n",
       " -73.355445238987684,\n",
       " -73.355445238987684,\n",
       " -73.20653172454962,\n",
       " -73.20653172454962,\n",
       " -73.180037515897268,\n",
       " -73.180037515897268,\n",
       " -73.043132692377014,\n",
       " -73.043132692377014,\n",
       " -73.01058344895408,\n",
       " -73.01058344895408,\n",
       " -72.967604824302541,\n",
       " -72.967604824302541,\n",
       " -72.959561175021975,\n",
       " -72.959561175021975,\n",
       " -72.784548335158306,\n",
       " -72.784548335158306,\n",
       " -72.711481332431475,\n",
       " -72.711481332431475,\n",
       " -72.503923348264451,\n",
       " -72.503923348264451,\n",
       " -72.484141932247354,\n",
       " -72.484141932247354,\n",
       " -72.39987859342682,\n",
       " -72.39987859342682,\n",
       " -72.155165149562649,\n",
       " -72.155165149562649,\n",
       " -72.134248321213036,\n",
       " -72.134248321213036,\n",
       " -71.569638763771394,\n",
       " -71.569638763771394,\n",
       " -71.468922807074051,\n",
       " -71.468922807074051,\n",
       " -71.070939402932197,\n",
       " -71.070939402932197,\n",
       " -71.060482177678736,\n",
       " -71.060482177678736,\n",
       " -70.931657810581811,\n",
       " -70.931657810581811,\n",
       " -70.763881702508357,\n",
       " -70.763881702508357,\n",
       " -70.646433687112676,\n",
       " -70.646433687112676,\n",
       " -70.641520814355161,\n",
       " -70.641520814355161,\n",
       " -70.472438632153569,\n",
       " -70.472438632153569,\n",
       " -70.441829211326962,\n",
       " -70.441829211326962,\n",
       " -70.310581401077044,\n",
       " -70.310581401077044,\n",
       " -70.271370795258903,\n",
       " -70.271370795258903,\n",
       " -70.203765844655337,\n",
       " -70.203765844655337,\n",
       " -69.787210171840655,\n",
       " -69.787210171840655,\n",
       " -69.748832477227182,\n",
       " -69.748832477227182,\n",
       " -69.731631081286238,\n",
       " -69.731631081286238,\n",
       " -69.594332813754306,\n",
       " -69.594332813754306,\n",
       " -69.557508558225479,\n",
       " -69.557508558225479,\n",
       " -69.498977252895202,\n",
       " -69.498977252895202,\n",
       " -69.441973207464088,\n",
       " -69.441973207464088,\n",
       " -69.09118931666579,\n",
       " -69.09118931666579,\n",
       " -68.851582296847042,\n",
       " -68.851582296847042,\n",
       " -68.774120036102858,\n",
       " -68.774120036102858,\n",
       " -68.533250760036282,\n",
       " -68.533250760036282,\n",
       " -68.484715666953733,\n",
       " -68.484715666953733,\n",
       " -68.342658882544981,\n",
       " -68.342658882544981,\n",
       " -68.335876159024778,\n",
       " -68.335876159024778,\n",
       " -68.321931487337338,\n",
       " -68.321931487337338,\n",
       " -68.153836431296583,\n",
       " -68.153836431296583,\n",
       " -68.086369729230654,\n",
       " -68.086369729230654,\n",
       " -68.080464649144659,\n",
       " -68.080464649144659,\n",
       " -67.974033580169461,\n",
       " -67.974033580169461,\n",
       " -67.911545271341183,\n",
       " -67.911545271341183,\n",
       " -67.881912878035493,\n",
       " -67.881912878035493,\n",
       " -67.838058537625869,\n",
       " -67.838058537625869,\n",
       " -67.822662046222604,\n",
       " -67.822662046222604,\n",
       " -67.800958119714778,\n",
       " -67.800958119714778,\n",
       " -67.796637285081346,\n",
       " -67.796637285081346,\n",
       " -67.761885727941774,\n",
       " -67.761885727941774,\n",
       " -67.528120536162334,\n",
       " -67.528120536162334,\n",
       " -67.42600052567083,\n",
       " -67.42600052567083,\n",
       " -67.328109792800731,\n",
       " -67.328109792800731,\n",
       " -67.082909658518318,\n",
       " -67.082909658518318,\n",
       " -66.629447241178156,\n",
       " -66.629447241178156,\n",
       " -66.466481395486625,\n",
       " -66.466481395486625,\n",
       " -66.453997012504928,\n",
       " -66.453997012504928,\n",
       " -66.303625844518876,\n",
       " -66.303625844518876,\n",
       " -66.189002326243653,\n",
       " -66.189002326243653,\n",
       " -66.169017428343182,\n",
       " -66.169017428343182,\n",
       " -66.158189183268902,\n",
       " -66.158189183268902,\n",
       " -66.131615857504073,\n",
       " -66.131615857504073,\n",
       " -65.988463015969899,\n",
       " -65.988463015969899,\n",
       " -65.787813759666562,\n",
       " -65.787813759666562,\n",
       " -65.781792333800908,\n",
       " -65.781792333800908,\n",
       " -65.745380194910211,\n",
       " -65.745380194910211,\n",
       " -65.699213368689797,\n",
       " -65.699213368689797,\n",
       " -65.694868912138901,\n",
       " -65.694868912138901,\n",
       " -65.65453520041919,\n",
       " -65.65453520041919,\n",
       " -65.620164143468074,\n",
       " -65.620164143468074,\n",
       " -65.609503420850729,\n",
       " -65.609503420850729,\n",
       " -65.608397841574458,\n",
       " -65.608397841574458,\n",
       " -65.599866713188291,\n",
       " -65.599866713188291,\n",
       " -65.435318992875352,\n",
       " -65.435318992875352,\n",
       " -65.307485601320906,\n",
       " -65.307485601320906,\n",
       " -65.302418101664458,\n",
       " -65.302418101664458,\n",
       " -65.141998415150979,\n",
       " -65.141998415150979,\n",
       " -65.133467135980652,\n",
       " -65.133467135980652,\n",
       " -64.850232087987692,\n",
       " -64.850232087987692,\n",
       " -64.741671368661585,\n",
       " -64.741671368661585,\n",
       " -64.609959387131795,\n",
       " -64.609959387131795,\n",
       " -64.182111101074298,\n",
       " -64.182111101074298,\n",
       " -64.165866571304761,\n",
       " -64.165866571304761,\n",
       " -64.028353370392935,\n",
       " -64.028353370392935,\n",
       " -63.970480438285321,\n",
       " -63.970480438285321,\n",
       " -63.895254087343851,\n",
       " -63.895254087343851,\n",
       " -63.83879731383734,\n",
       " -63.83879731383734,\n",
       " -63.817430238438106,\n",
       " -63.817430238438106,\n",
       " -63.592876340282288,\n",
       " -63.592876340282288,\n",
       " -63.566008026629447,\n",
       " -63.566008026629447,\n",
       " -63.493630146297306,\n",
       " -63.493630146297306,\n",
       " -63.477555546309823,\n",
       " -63.477555546309823,\n",
       " -63.428121000218617,\n",
       " -63.428121000218617,\n",
       " -63.416494795058746,\n",
       " -63.416494795058746,\n",
       " -63.364556001379782,\n",
       " -63.364556001379782,\n",
       " -63.337233710082501,\n",
       " -63.337233710082501,\n",
       " -63.138995582051798,\n",
       " -63.138995582051798,\n",
       " -63.048884160619274,\n",
       " -63.048884160619274,\n",
       " -63.039185772758486,\n",
       " -63.039185772758486,\n",
       " -63.004844594130681,\n",
       " -63.004844594130681,\n",
       " -63.002285053974418,\n",
       " -63.002285053974418,\n",
       " -62.942427573998138,\n",
       " -62.942427573998138,\n",
       " -62.929995001632932,\n",
       " -62.929995001632932,\n",
       " -62.902136694543998,\n",
       " -62.902136694543998,\n",
       " -62.88086251060016,\n",
       " -62.88086251060016,\n",
       " -62.738690526654736,\n",
       " -62.738690526654736,\n",
       " -62.703157512425044,\n",
       " -62.703157512425044,\n",
       " -62.579333248350956,\n",
       " -62.579333248350956,\n",
       " -62.33158204163977,\n",
       " -62.33158204163977,\n",
       " -61.81011594414241,\n",
       " -61.81011594414241,\n",
       " -61.745505688954474,\n",
       " -61.745505688954474,\n",
       " -61.725428747815158,\n",
       " -61.725428747815158,\n",
       " -61.717644599735237,\n",
       " -61.717644599735237,\n",
       " -61.5982514447943,\n",
       " -61.5982514447943,\n",
       " -61.527750971312358,\n",
       " -61.527750971312358,\n",
       " -61.484186527557249,\n",
       " -61.484186527557249,\n",
       " -61.475421035584581,\n",
       " -61.475421035584581,\n",
       " -61.459143683681631,\n",
       " -61.459143683681631,\n",
       " -61.451915158323473,\n",
       " -61.451915158323473,\n",
       " -61.445241675691129,\n",
       " -61.445241675691129,\n",
       " -61.441388800878364,\n",
       " -61.441388800878364,\n",
       " -61.349127753394114,\n",
       " -61.349127753394114,\n",
       " -61.34099223691404,\n",
       " -61.34099223691404,\n",
       " -61.304654500706214,\n",
       " -61.304654500706214,\n",
       " -61.183132432550515,\n",
       " -61.183132432550515,\n",
       " -61.168029090365373,\n",
       " -61.168029090365373,\n",
       " -61.159997814426958,\n",
       " -61.159997814426958,\n",
       " -61.138816209238215,\n",
       " -61.138816209238215,\n",
       " -61.1355396491645,\n",
       " -61.1355396491645,\n",
       " -60.909476880730871,\n",
       " -60.909476880730871,\n",
       " -60.904791004455433,\n",
       " -60.904791004455433,\n",
       " -60.629379547383259,\n",
       " -60.629379547383259,\n",
       " -60.518632406120886,\n",
       " -60.518632406120886,\n",
       " -60.503741237963155,\n",
       " -60.503741237963155,\n",
       " -60.481757695208799,\n",
       " -60.481757695208799,\n",
       " -60.413345981320113,\n",
       " -60.413345981320113,\n",
       " -60.408307789480801,\n",
       " -60.408307789480801,\n",
       " -60.316031867832329,\n",
       " -60.316031867832329,\n",
       " -60.244243230918059,\n",
       " -60.244243230918059,\n",
       " -60.226987575301834,\n",
       " -60.226987575301834,\n",
       " -60.183577397275322,\n",
       " -60.183577397275322,\n",
       " -60.116748623524515,\n",
       " -60.116748623524515,\n",
       " -60.090896182970077,\n",
       " -60.090896182970077,\n",
       " -60.030962783185366,\n",
       " -60.030962783185366,\n",
       " -60.018849137986358,\n",
       " -60.018849137986358,\n",
       " -59.944560711421197,\n",
       " -59.944560711421197,\n",
       " -59.876257220155246,\n",
       " -59.876257220155246,\n",
       " -59.844071977046276,\n",
       " -59.844071977046276,\n",
       " -59.779091538634916,\n",
       " -59.779091538634916,\n",
       " -59.729720604445859,\n",
       " -59.729720604445859,\n",
       " -59.63062847033428,\n",
       " -59.63062847033428,\n",
       " -59.589571145686051,\n",
       " -59.589571145686051,\n",
       " -59.519182856475041,\n",
       " -59.519182856475041,\n",
       " -59.484915574701496,\n",
       " -59.484915574701496,\n",
       " -59.442477720076646,\n",
       " -59.442477720076646,\n",
       " -59.361563761904769,\n",
       " -59.361563761904769,\n",
       " -59.360051558041548,\n",
       " -59.360051558041548,\n",
       " -59.353614750073469,\n",
       " -59.353614750073469,\n",
       " -59.287037199308415,\n",
       " -59.287037199308415,\n",
       " -59.256786267516709,\n",
       " -59.256786267516709,\n",
       " -59.152896370861811,\n",
       " -59.152896370861811,\n",
       " -59.151867429440799,\n",
       " -59.151867429440799,\n",
       " -59.138943184028413,\n",
       " -59.138943184028413,\n",
       " -59.066033173371508,\n",
       " -59.066033173371508,\n",
       " -59.000310950981678,\n",
       " -59.000310950981678,\n",
       " -58.994454710147004,\n",
       " -58.994454710147004,\n",
       " -58.676092940276938,\n",
       " -58.676092940276938,\n",
       " -58.645659799440935,\n",
       " -58.645659799440935,\n",
       " -58.587513704915807,\n",
       " -58.587513704915807,\n",
       " -58.575212629673288,\n",
       " -58.575212629673288,\n",
       " -58.535456622332326,\n",
       " -58.535456622332326,\n",
       " -58.475768056166203,\n",
       " -58.475768056166203,\n",
       " -58.46069783830589,\n",
       " -58.46069783830589,\n",
       " -58.432705337102874,\n",
       " -58.432705337102874,\n",
       " -58.428319754512103,\n",
       " -58.428319754512103,\n",
       " -58.411527022130286,\n",
       " -58.411527022130286,\n",
       " -58.376847592883557,\n",
       " -58.376847592883557,\n",
       " -58.356325321768026,\n",
       " -58.356325321768026,\n",
       " -58.349913089814052,\n",
       " -58.349913089814052,\n",
       " -58.322684712108945,\n",
       " -58.322684712108945,\n",
       " -58.251688661369585,\n",
       " -58.251688661369585,\n",
       " -58.247704567290661,\n",
       " -58.247704567290661,\n",
       " -58.243082929973369,\n",
       " -58.243082929973369,\n",
       " -58.021286777535458,\n",
       " -58.021286777535458,\n",
       " -57.970477708416048,\n",
       " -57.970477708416048,\n",
       " -57.937057920435919,\n",
       " -57.937057920435919,\n",
       " -57.924577206128305,\n",
       " -57.924577206128305,\n",
       " -57.906075487215318,\n",
       " -57.906075487215318,\n",
       " -57.875198191789757,\n",
       " -57.875198191789757,\n",
       " -57.857050208313133,\n",
       " -57.857050208313133,\n",
       " -57.79861290547732,\n",
       " -57.79861290547732,\n",
       " -57.776540291455284,\n",
       " -57.776540291455284,\n",
       " -57.771865593047025,\n",
       " -57.771865593047025,\n",
       " -57.749386434038627,\n",
       " -57.749386434038627,\n",
       " -57.730389876511843,\n",
       " -57.730389876511843,\n",
       " -57.727073280840408,\n",
       " -57.727073280840408,\n",
       " -57.71844028905285,\n",
       " -57.71844028905285,\n",
       " -57.698381625779376,\n",
       " -57.698381625779376,\n",
       " -57.686591271858696,\n",
       " -57.686591271858696,\n",
       " -57.657398847372789,\n",
       " -57.657398847372789,\n",
       " -57.625247425503645,\n",
       " -57.625247425503645,\n",
       " -57.618199462683535,\n",
       " -57.618199462683535,\n",
       " -57.610307535243578,\n",
       " -57.610307535243578,\n",
       " -57.589696306599578,\n",
       " -57.589696306599578,\n",
       " -57.558654530543613,\n",
       " -57.558654530543613,\n",
       " -57.492010081408019,\n",
       " -57.492010081408019,\n",
       " -57.41430136202537,\n",
       " -57.41430136202537,\n",
       " -57.344347499585879,\n",
       " -57.344347499585879,\n",
       " -57.304855830644335,\n",
       " -57.304855830644335,\n",
       " -57.273667232582405,\n",
       " -57.273667232582405,\n",
       " -57.270280178751442,\n",
       " -57.270280178751442,\n",
       " -57.254138268886834,\n",
       " -57.254138268886834,\n",
       " -57.242403565146091,\n",
       " -57.242403565146091,\n",
       " -57.237998386795823,\n",
       " -57.237998386795823,\n",
       " -57.225983109711365,\n",
       " -57.225983109711365,\n",
       " -57.175477832058178,\n",
       " -57.175477832058178,\n",
       " -57.172043918898112,\n",
       " -57.172043918898112,\n",
       " -57.131115086408897,\n",
       " -57.131115086408897,\n",
       " -57.113768227795298,\n",
       " -57.113768227795298,\n",
       " -56.933037143977835,\n",
       " -56.933037143977835,\n",
       " -56.917845383030333,\n",
       " -56.917845383030333,\n",
       " -56.844642575540647,\n",
       " -56.844642575540647,\n",
       " -56.833683960045406,\n",
       " -56.833683960045406,\n",
       " -56.833606305718192,\n",
       " -56.833606305718192,\n",
       " -56.80165845984741,\n",
       " -56.80165845984741,\n",
       " -56.789932555297028,\n",
       " -56.789932555297028,\n",
       " -56.734421375037435,\n",
       " -56.734421375037435,\n",
       " -56.72638537174651,\n",
       " -56.72638537174651,\n",
       " -56.711518363095976,\n",
       " -56.711518363095976,\n",
       " -56.678416646240194,\n",
       " -56.678416646240194,\n",
       " -56.64308595641198,\n",
       " -56.64308595641198,\n",
       " -56.577462726885827,\n",
       " -56.577462726885827,\n",
       " -56.572538618315491,\n",
       " -56.572538618315491,\n",
       " -56.523857601160259,\n",
       " -56.523857601160259,\n",
       " -56.479141706332953,\n",
       " -56.479141706332953,\n",
       " -56.442605123306116,\n",
       " -56.442605123306116,\n",
       " -56.434450606123683,\n",
       " -56.434450606123683,\n",
       " -56.41462714660625,\n",
       " -56.41462714660625,\n",
       " -56.393595389342835,\n",
       " -56.393595389342835,\n",
       " -56.374071096573005,\n",
       " -56.374071096573005,\n",
       " -56.302568756912592,\n",
       " -56.302568756912592,\n",
       " -56.297988746184238,\n",
       " -56.297988746184238,\n",
       " -56.240040535160624,\n",
       " -56.240040535160624,\n",
       " -56.233702794459525,\n",
       " -56.233702794459525,\n",
       " -56.128967561660161,\n",
       " -56.128967561660161,\n",
       " -56.044502153901455,\n",
       " -56.044502153901455,\n",
       " -56.01879619262715,\n",
       " -56.01879619262715,\n",
       " -56.010199640737973,\n",
       " -56.010199640737973,\n",
       " -55.95929397803355,\n",
       " -55.95929397803355,\n",
       " -55.839890352526439,\n",
       " -55.839890352526439,\n",
       " -55.822120379603895,\n",
       " -55.822120379603895,\n",
       " -55.811124241813587,\n",
       " -55.811124241813587,\n",
       " -55.690063543413046,\n",
       " -55.690063543413046,\n",
       " -55.601430645418233,\n",
       " -55.601430645418233,\n",
       " -55.496780690742625,\n",
       " -55.496780690742625,\n",
       " -55.41110896393208,\n",
       " -55.41110896393208,\n",
       " -55.312125935405376,\n",
       " -55.312125935405376,\n",
       " -55.303707382216579,\n",
       " -55.303707382216579,\n",
       " -55.292587064317722,\n",
       " -55.292587064317722,\n",
       " -55.28786979580331,\n",
       " -55.28786979580331,\n",
       " -55.2196220304957,\n",
       " -55.2196220304957,\n",
       " -55.192360879819425,\n",
       " -55.192360879819425,\n",
       " -55.152272437967021,\n",
       " -55.152272437967021,\n",
       " -55.137054683565964,\n",
       " -55.137054683565964,\n",
       " -55.112830427617908,\n",
       " -55.112830427617908,\n",
       " -55.095264491841036,\n",
       " -55.095264491841036,\n",
       " -55.020206029509296,\n",
       " -55.020206029509296,\n",
       " -54.987848719217745,\n",
       " -54.987848719217745,\n",
       " -54.904000734396917,\n",
       " -54.904000734396917,\n",
       " -54.87744513873168,\n",
       " -54.87744513873168,\n",
       " -54.875007419648277,\n",
       " -54.875007419648277,\n",
       " -54.865781759211579,\n",
       " -54.865781759211579,\n",
       " -54.785409479787589,\n",
       " -54.785409479787589,\n",
       " -54.759585215631553,\n",
       " -54.759585215631553,\n",
       " -54.739991076604355,\n",
       " -54.739991076604355,\n",
       " -54.735462026863722,\n",
       " -54.735462026863722,\n",
       " -54.704616025506944,\n",
       " -54.704616025506944,\n",
       " -54.686212570659883,\n",
       " -54.686212570659883,\n",
       " -54.684019693634312,\n",
       " -54.684019693634312,\n",
       " -54.671112832156247,\n",
       " -54.671112832156247,\n",
       " -54.558666910409748,\n",
       " -54.558666910409748,\n",
       " -54.545237069859432,\n",
       " -54.545237069859432,\n",
       " -54.487922834564714,\n",
       " -54.487922834564714,\n",
       " -54.473629134337216,\n",
       " -54.473629134337216,\n",
       " -54.470083714742529,\n",
       " -54.470083714742529,\n",
       " -54.456093818083318,\n",
       " -54.456093818083318,\n",
       " -54.445244323788614,\n",
       " -54.445244323788614,\n",
       " -54.418379375960782,\n",
       " -54.418379375960782,\n",
       " -54.412614072941359,\n",
       " -54.412614072941359,\n",
       " -54.394835246554074,\n",
       " -54.394835246554074,\n",
       " -54.282348269643222,\n",
       " -54.282348269643222,\n",
       " -54.149217201159118,\n",
       " -54.149217201159118,\n",
       " -54.1437601755241,\n",
       " -54.1437601755241,\n",
       " -54.12674016373694,\n",
       " -54.12674016373694,\n",
       " -54.071688884190607,\n",
       " -54.071688884190607,\n",
       " -54.061468683429879,\n",
       " -54.061468683429879,\n",
       " -54.023563138251617,\n",
       " -54.023563138251617,\n",
       " -53.938479646695335,\n",
       " -53.938479646695335,\n",
       " -53.902416030212351,\n",
       " -53.902416030212351,\n",
       " -53.864690304185309,\n",
       " -53.864690304185309,\n",
       " -53.772517793556666,\n",
       " -53.772517793556666,\n",
       " -53.707608314369445,\n",
       " -53.707608314369445,\n",
       " -53.651916282193284,\n",
       " -53.651916282193284,\n",
       " -53.638890737004211,\n",
       " -53.638890737004211,\n",
       " -53.61888260963682,\n",
       " -53.61888260963682,\n",
       " -53.592000261843012,\n",
       " -53.592000261843012,\n",
       " -53.586118196232235,\n",
       " -53.586118196232235,\n",
       " -53.572337462651021,\n",
       " -53.572337462651021,\n",
       " -53.566811614512453,\n",
       " -53.566811614512453,\n",
       " -53.533684013790548,\n",
       " -53.533684013790548,\n",
       " -53.521463151591234,\n",
       " -53.521463151591234,\n",
       " -53.513610298673441,\n",
       " -53.513610298673441,\n",
       " -53.497042005536287,\n",
       " -53.497042005536287,\n",
       " -53.478894975066041,\n",
       " -53.478894975066041,\n",
       " -53.398558342647142,\n",
       " -53.398558342647142,\n",
       " -53.360835656986147,\n",
       " -53.360835656986147,\n",
       " -53.335366185519085,\n",
       " -53.335366185519085,\n",
       " -53.325732267880078,\n",
       " -53.325732267880078,\n",
       " -53.318435373075815,\n",
       " -53.318435373075815,\n",
       " -53.269693801153466,\n",
       " -53.269693801153466,\n",
       " -53.249596602737647,\n",
       " -53.249596602737647,\n",
       " -53.240346155494102,\n",
       " -53.240346155494102,\n",
       " -53.234637954977138,\n",
       " -53.234637954977138,\n",
       " -53.229933693290796,\n",
       " -53.229933693290796,\n",
       " -53.210669382777752,\n",
       " -53.210669382777752,\n",
       " -53.118080917391133,\n",
       " -53.118080917391133,\n",
       " -53.077508300810493,\n",
       " -53.077508300810493,\n",
       " -53.022396396577925,\n",
       " -53.022396396577925,\n",
       " -53.022074191019968,\n",
       " -53.022074191019968,\n",
       " -53.014297865692484,\n",
       " -53.014297865692484,\n",
       " -52.96245308956059,\n",
       " -52.96245308956059,\n",
       " -52.932594136970216,\n",
       " -52.932594136970216,\n",
       " -52.912996130672013,\n",
       " -52.912996130672013,\n",
       " -52.897425511046009,\n",
       " -52.897425511046009,\n",
       " -52.894958250360929,\n",
       " -52.894958250360929,\n",
       " -52.889652654346932,\n",
       " -52.889652654346932,\n",
       " -52.875213969279592,\n",
       " -52.875213969279592,\n",
       " -52.858400403077098,\n",
       " -52.858400403077098,\n",
       " -52.844660018386904,\n",
       " -52.844660018386904,\n",
       " -52.790193129673334,\n",
       " -52.790193129673334,\n",
       " -52.788528684899781,\n",
       " -52.788528684899781,\n",
       " -52.737921128415401,\n",
       " -52.737921128415401,\n",
       " -52.701338695508561,\n",
       " -52.701338695508561,\n",
       " -52.620535365132291,\n",
       " -52.620535365132291,\n",
       " -52.611765569745678,\n",
       " -52.611765569745678,\n",
       " -52.609175840547906,\n",
       " -52.609175840547906,\n",
       " -52.552629737512255,\n",
       " -52.552629737512255,\n",
       " -52.525710465389444,\n",
       " -52.525710465389444,\n",
       " -52.515614738975529,\n",
       " -52.515614738975529,\n",
       " -52.499572570634953,\n",
       " -52.499572570634953,\n",
       " -52.423970843724888,\n",
       " -52.423970843724888,\n",
       " -52.408676417948293,\n",
       " -52.408676417948293,\n",
       " -52.386970199153509,\n",
       " -52.386970199153509,\n",
       " -52.386589052610908,\n",
       " -52.386589052610908,\n",
       " -52.370067701711768,\n",
       " -52.370067701711768,\n",
       " -52.358430466118598,\n",
       " -52.358430466118598,\n",
       " -52.312923157289916,\n",
       " -52.312923157289916,\n",
       " -52.305004136424692,\n",
       " -52.305004136424692,\n",
       " -52.28437717111386,\n",
       " -52.28437717111386,\n",
       " -52.12815772365721,\n",
       " -52.12815772365721,\n",
       " -52.056642200463443,\n",
       " -52.056642200463443,\n",
       " -52.029595366293698,\n",
       " -52.029595366293698,\n",
       " -51.973046803045804,\n",
       " -51.973046803045804,\n",
       " -51.94965602330209,\n",
       " -51.94965602330209,\n",
       " -51.944050934287844,\n",
       " -51.944050934287844,\n",
       " -51.902169990315656,\n",
       " -51.902169990315656,\n",
       " -51.856023978077964,\n",
       " -51.856023978077964,\n",
       " -51.852192334035564,\n",
       " -51.852192334035564,\n",
       " -51.848366639091829,\n",
       " -51.848366639091829,\n",
       " -51.820858202327265,\n",
       " -51.820858202327265,\n",
       " -51.807472690007913,\n",
       " -51.807472690007913,\n",
       " -51.784458638530552,\n",
       " -51.784458638530552,\n",
       " -51.703811301429859,\n",
       " -51.703811301429859,\n",
       " -51.691313204097909,\n",
       " -51.691313204097909,\n",
       " -51.620911304722092,\n",
       " -51.620911304722092,\n",
       " -51.594704821819043,\n",
       " -51.594704821819043,\n",
       " -51.590329565779115,\n",
       " -51.590329565779115,\n",
       " -51.582950466039968,\n",
       " -51.582950466039968,\n",
       " -51.560547965932201,\n",
       " -51.560547965932201,\n",
       " -51.553600331861723,\n",
       " -51.553600331861723,\n",
       " -51.529989488477391,\n",
       " -51.529989488477391,\n",
       " -51.51844229221043,\n",
       " -51.51844229221043,\n",
       " -51.514810533767069,\n",
       " -51.514810533767069,\n",
       " -51.417362795055702,\n",
       " -51.417362795055702,\n",
       " ...]"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "logs = np.log(non_zero_val)\n",
    "sorted(logs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "logs_df = pd.DataFrame(sorted(logs))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f1146d39320>"
      ]
     },
     "execution_count": 108,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXwAAAD3CAYAAAAe5+9lAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1114512358>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "logs_df.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7f11ee77f400>"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAD3CAYAAADi8sSvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkhJiqEtSQgx1SUqIoS5JCfn/K5zyVU0Ycq4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f11ee7bd240>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Taking a look at the embedding again\n",
    "df = pd.DataFrame(review_embedding, columns=['x', 'y'])\n",
    "df['labels'] = reviews['sentiment']\n",
    "plt.scatter(x=df['x'], y=df['y'], c=df['labels'], cmap='tab10')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Looking for the outlying ones"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>x</th>\n",
       "      <th>y</th>\n",
       "      <th>labels</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1751</th>\n",
       "      <td>98.998522</td>\n",
       "      <td>0.007130</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2483</th>\n",
       "      <td>0.011736</td>\n",
       "      <td>98.216747</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              x          y  labels\n",
       "1751  98.998522   0.007130       1\n",
       "2483   0.011736  98.216747       1"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[(df.x > 80) | (df.y > 80)] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "outlier_row_1 = graph[1751, :]\n",
    "outlier_col_1 = graph[:, 1751]\n",
    "\n",
    "outlier_row_2 = graph[2483, :]\n",
    "outlier_col_2 = graph[:, 2483]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def get_logs_of_non_zero(v):\n",
    "    i, j, k = find(v)\n",
    "    return np.log(k)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "row_1_logs = get_logs_of_non_zero(outlier_row_1)\n",
    "col_1_logs = get_logs_of_non_zero(outlier_col_1)\n",
    "\n",
    "row_2_logs = get_logs_of_non_zero(outlier_row_2)\n",
    "col_2_logs = get_logs_of_non_zero(outlier_col_2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Let's look at the outlying row distributions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[-121.72089185206474,\n",
       " -120.62853971293048,\n",
       " -118.06674479569836,\n",
       " -115.82457962375778,\n",
       " -114.40917546049887,\n",
       " -112.74276101187854,\n",
       " -110.80671431346163,\n",
       " -109.63008147114049,\n",
       " -109.4795346279987,\n",
       " -105.53656502373077]"
      ]
     },
     "execution_count": 103,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sorted(row_1_logs)[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[-182.76703723076869,\n",
       " -155.61499637503186,\n",
       " -111.45909513242631,\n",
       " -109.84968350614497,\n",
       " -109.5757808450868,\n",
       " -109.11587072471065,\n",
       " -104.51705211744323,\n",
       " -99.757132353921378,\n",
       " -96.433267756684558,\n",
       " -91.417401193614552]"
      ]
     },
     "execution_count": 102,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sorted(row_2_logs)[:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Picking random rows"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Top 10 values of row 13294\n",
      "[-17.647719931089348, -16.719463363896093, -16.399787350007774, -16.394554000585664, -16.061055366323178, -15.983542014369188, -15.589802920356556, -15.393751995816006, -14.819123661464554, -14.794492960408853]\n",
      "Top 10 values of row 23701\n",
      "[-22.106675481114149, -21.742645096840938, -17.482367490025009, -16.152028930589609, -15.806665220384867, -15.76284545822922, -14.873900722291518, -14.361575454195519, -13.732103081932046, -13.499198209696276]\n",
      "Top 10 values of row 5332\n",
      "[-20.705514882766529, -14.999716820855555, -14.530156560886741, -14.379879745197625, -14.309015863457779, -12.249114475336517, -11.757046953190208, -11.627936364645517, -11.171694667042717, -10.74493496434053]\n",
      "Top 10 values of row 7676\n",
      "[-16.797223733069799, -16.646816333393961, -15.684379186385122, -15.590805657698843, -14.759392629305275, -13.076956007953662, -12.989904224158789, -12.703244285086072, -12.099902290658584, -12.07369052066027]\n",
      "Top 10 values of row 18152\n",
      "[-26.193783734737433, -25.424027989644497, -24.819670305174121, -22.717433473476753, -21.278694657452984, -19.368103373854701, -19.30461318686379, -18.766114483224516, -18.702503117645893, -18.600732206758821]\n",
      "Top 10 values of row 14543\n",
      "[-74.501304468463843, -66.158189183268902, -63.592876340282288, -54.759585215631553, -52.620535365132291, -50.48873758441853, -45.143647093858213, -43.739173950626707, -42.442731650367712, -41.500479069972506]\n",
      "Top 10 values of row 16654\n",
      "[-15.635096589433596, -15.475807147690626, -13.498683051242342, -13.358579875530419, -13.183936502944992, -13.080866476422175, -13.052011349784456, -12.273627660047641, -12.113664399360511, -12.047982556270497]\n",
      "Top 10 values of row 19615\n",
      "[-22.323460813882168, -21.057143847697553, -18.611793953922767, -17.915833469246348, -17.773764333843506, -17.296841583905032, -15.54798702969107, -15.055348492635334, -14.704565367561596, -14.701259490881494]\n",
      "Top 10 values of row 21817\n",
      "[-14.230788686918654, -13.500702674799186, -12.445416434539519, -11.861731071692486, -11.601821937440524, -11.343292859232109, -11.021707099589365, -10.627239246491868, -10.481665716548541, -10.481211209726547]\n",
      "Top 10 values of row 6043\n",
      "[-19.050080201429918, -19.03192827492143, -15.253643762746886, -14.553359412623395, -13.620873530395212, -12.991802606816639, -12.831354818941378, -12.198479173062744, -12.170451990917226, -12.05028620539135]\n"
     ]
    }
   ],
   "source": [
    "rand_ints = np.random.randint(graph.shape[0], size=10)\n",
    "for r in rand_ints:\n",
    "    row_logs = get_logs_of_non_zero(graph[r, :])\n",
    "    col_logs = get_logs_of_non_zero(graph[:, r])\n",
    "    \n",
    "    print('Top 10 values of row {}'.format(r))\n",
    "    print(sorted(row_logs)[:10])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So it appears the logs of the outlying rows are much bigger than normal. Not sure what this means"
   ]
  },
  {
   "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": 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": []
  }
 ],
 "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}