sentiment.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sentiment Analysis of #FirstAmendment Tweets"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The goal of this project is to explore the way that Twitter users define and invoke the first amendment. Who are the \"defenders of free speech\" on the web and what are they saying? \n",
    "\n",
    "I'm interested in grouping these people by political leaning. Do liberals and conservatives participate differently in debates on free speech?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import nltk;\n",
    "from textblob import TextBlob\n",
    "import pandas as pd\n",
    "import numpy\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data\n",
    "I've pulled 450 tweets tagged with #firstamendment and have stored them in a csv file. Each row contains the timestamp, tweet text, number of retweets and likes, and whether or not the user follows Donald Trump and/or Hillary Clinton.\n",
    "\n",
    "This last metric is being used as a loose method of identifying political leaning. I understand that following either of these figures is not a comprehensive method of predicting political leaning but have chosen this approach for early analysis. I plan to refine this method as I continue. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('tweets-11-11.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>created at</th>\n",
       "      <th>text</th>\n",
       "      <th>retweets</th>\n",
       "      <th>likes</th>\n",
       "      <th>follows trump</th>\n",
       "      <th>follows clinton</th>\n",
       "      <th>sentiment polarity</th>\n",
       "      <th>sentiment subjectivity</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2019-11-11 16:25:30</td>\n",
       "      <td>b'RT @kleetampa: Dear @realDonaldTrump \\nWheth...</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2019-11-11 16:24:32</td>\n",
       "      <td>b'RT @DevinCow: Just a reminder that there is ...</td>\n",
       "      <td>2369</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2019-11-11 16:12:05</td>\n",
       "      <td>b'RT @DevinCow: Just a reminder that there is ...</td>\n",
       "      <td>2369</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2019-11-11 16:11:40</td>\n",
       "      <td>b'This was absolutely glorious. As best an arg...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>0.600000</td>\n",
       "      <td>0.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2019-11-11 16:11:38</td>\n",
       "      <td>b'RT @Rose_glow1: Trump let them in the front ...</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>445</th>\n",
       "      <td>2019-11-10 08:37:09</td>\n",
       "      <td>b'RT @DevinCow: Just a reminder that there is ...</td>\n",
       "      <td>2372</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>446</th>\n",
       "      <td>2019-11-10 08:25:16</td>\n",
       "      <td>b'RT @DevinCow: Just a reminder that there is ...</td>\n",
       "      <td>2372</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>447</th>\n",
       "      <td>2019-11-10 07:58:29</td>\n",
       "      <td>b'RT @politicallands1: We are currently being ...</td>\n",
       "      <td>18</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>0.112121</td>\n",
       "      <td>0.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>448</th>\n",
       "      <td>2019-11-10 07:52:26</td>\n",
       "      <td>b'RT @DevinCow: Just a reminder that there is ...</td>\n",
       "      <td>2372</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>449</th>\n",
       "      <td>2019-11-10 07:33:07</td>\n",
       "      <td>b'RT @TheCiscoKid2: #ResistLouder #FirstAmendm...</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>450 rows × 8 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              created at                                               text  \\\n",
       "0    2019-11-11 16:25:30  b'RT @kleetampa: Dear @realDonaldTrump \\nWheth...   \n",
       "1    2019-11-11 16:24:32  b'RT @DevinCow: Just a reminder that there is ...   \n",
       "2    2019-11-11 16:12:05  b'RT @DevinCow: Just a reminder that there is ...   \n",
       "3    2019-11-11 16:11:40  b'This was absolutely glorious. As best an arg...   \n",
       "4    2019-11-11 16:11:38  b'RT @Rose_glow1: Trump let them in the front ...   \n",
       "..                   ...                                                ...   \n",
       "445  2019-11-10 08:37:09  b'RT @DevinCow: Just a reminder that there is ...   \n",
       "446  2019-11-10 08:25:16  b'RT @DevinCow: Just a reminder that there is ...   \n",
       "447  2019-11-10 07:58:29  b'RT @politicallands1: We are currently being ...   \n",
       "448  2019-11-10 07:52:26  b'RT @DevinCow: Just a reminder that there is ...   \n",
       "449  2019-11-10 07:33:07  b'RT @TheCiscoKid2: #ResistLouder #FirstAmendm...   \n",
       "\n",
       "     retweets  likes  follows trump  follows clinton  sentiment polarity  \\\n",
       "0           2      0           True             True            0.200000   \n",
       "1        2369      0          False            False            0.200000   \n",
       "2        2369      0          False             True            0.200000   \n",
       "3           0      0          False            False            0.600000   \n",
       "4           3      0           True             True            0.000000   \n",
       "..        ...    ...            ...              ...                 ...   \n",
       "445      2372      0          False             True            0.200000   \n",
       "446      2372      0          False             True            0.200000   \n",
       "447        18      0          False             True            0.112121   \n",
       "448      2372      0          False             True            0.200000   \n",
       "449         3      0           True             True            0.000000   \n",
       "\n",
       "     sentiment subjectivity  \n",
       "0                       0.2  \n",
       "1                       0.2  \n",
       "2                       0.2  \n",
       "3                       0.6  \n",
       "4                       0.0  \n",
       "..                      ...  \n",
       "445                     0.2  \n",
       "446                     0.2  \n",
       "447                     0.6  \n",
       "448                     0.2  \n",
       "449                     0.0  \n",
       "\n",
       "[450 rows x 8 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sentiment_polarity = []\n",
    "sentiment_subjectivity = []\n",
    "\n",
    "for index, row in df.iterrows():\n",
    "    sentiment_polarity.append(TextBlob(row['text']).sentiment.polarity)\n",
    "    sentiment_subjectivity.append(TextBlob(row['text']).sentiment.subjectivity)\n",
    "\n",
    "df['sentiment polarity'] = sentiment_polarity\n",
    "df['sentiment subjectivity'] = sentiment_subjectivity\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Sentiment Analysis\n",
    "Using the TextBlob API for sentiment analysis, I processed each tweet by its sentiment polarity (how positive or negative it is) and subjectivity.\n",
    "\n",
    "Because the TextBlob model is not trained specifically for this data, I plan to continue exploring other sentiment analysis models that may be more appropriate for social media data. \n",
    "\n",
    "Lastly, I've assigned each tweet a color based on who they follow: red for Trump, blue for Clinton, purple for both and gray for neither."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>created at</th>\n",
       "      <th>text</th>\n",
       "      <th>retweets</th>\n",
       "      <th>likes</th>\n",
       "      <th>follows trump</th>\n",
       "      <th>follows clinton</th>\n",
       "      <th>sentiment polarity</th>\n",
       "      <th>sentiment subjectivity</th>\n",
       "      <th>color</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2019-11-11 16:25:30</td>\n",
       "      <td>b'RT @kleetampa: Dear @realDonaldTrump \\nWheth...</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.2</td>\n",
       "      <td>red</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2019-11-11 16:24:32</td>\n",
       "      <td>b'RT @DevinCow: Just a reminder that there is ...</td>\n",
       "      <td>2369</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.2</td>\n",
       "      <td>gray</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2019-11-11 16:12:05</td>\n",
       "      <td>b'RT @DevinCow: Just a reminder that there is ...</td>\n",
       "      <td>2369</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.2</td>\n",
       "      <td>blue</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2019-11-11 16:11:40</td>\n",
       "      <td>b'This was absolutely glorious. As best an arg...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>0.600000</td>\n",
       "      <td>0.6</td>\n",
       "      <td>gray</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2019-11-11 16:11:38</td>\n",
       "      <td>b'RT @Rose_glow1: Trump let them in the front ...</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>red</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>445</th>\n",
       "      <td>2019-11-10 08:37:09</td>\n",
       "      <td>b'RT @DevinCow: Just a reminder that there is ...</td>\n",
       "      <td>2372</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.2</td>\n",
       "      <td>blue</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>446</th>\n",
       "      <td>2019-11-10 08:25:16</td>\n",
       "      <td>b'RT @DevinCow: Just a reminder that there is ...</td>\n",
       "      <td>2372</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.2</td>\n",
       "      <td>blue</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>447</th>\n",
       "      <td>2019-11-10 07:58:29</td>\n",
       "      <td>b'RT @politicallands1: We are currently being ...</td>\n",
       "      <td>18</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>0.112121</td>\n",
       "      <td>0.6</td>\n",
       "      <td>blue</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>448</th>\n",
       "      <td>2019-11-10 07:52:26</td>\n",
       "      <td>b'RT @DevinCow: Just a reminder that there is ...</td>\n",
       "      <td>2372</td>\n",
       "      <td>0</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.2</td>\n",
       "      <td>blue</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>449</th>\n",
       "      <td>2019-11-10 07:33:07</td>\n",
       "      <td>b'RT @TheCiscoKid2: #ResistLouder #FirstAmendm...</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>red</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>450 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              created at                                               text  \\\n",
       "0    2019-11-11 16:25:30  b'RT @kleetampa: Dear @realDonaldTrump \\nWheth...   \n",
       "1    2019-11-11 16:24:32  b'RT @DevinCow: Just a reminder that there is ...   \n",
       "2    2019-11-11 16:12:05  b'RT @DevinCow: Just a reminder that there is ...   \n",
       "3    2019-11-11 16:11:40  b'This was absolutely glorious. As best an arg...   \n",
       "4    2019-11-11 16:11:38  b'RT @Rose_glow1: Trump let them in the front ...   \n",
       "..                   ...                                                ...   \n",
       "445  2019-11-10 08:37:09  b'RT @DevinCow: Just a reminder that there is ...   \n",
       "446  2019-11-10 08:25:16  b'RT @DevinCow: Just a reminder that there is ...   \n",
       "447  2019-11-10 07:58:29  b'RT @politicallands1: We are currently being ...   \n",
       "448  2019-11-10 07:52:26  b'RT @DevinCow: Just a reminder that there is ...   \n",
       "449  2019-11-10 07:33:07  b'RT @TheCiscoKid2: #ResistLouder #FirstAmendm...   \n",
       "\n",
       "     retweets  likes  follows trump  follows clinton  sentiment polarity  \\\n",
       "0           2      0           True             True            0.200000   \n",
       "1        2369      0          False            False            0.200000   \n",
       "2        2369      0          False             True            0.200000   \n",
       "3           0      0          False            False            0.600000   \n",
       "4           3      0           True             True            0.000000   \n",
       "..        ...    ...            ...              ...                 ...   \n",
       "445      2372      0          False             True            0.200000   \n",
       "446      2372      0          False             True            0.200000   \n",
       "447        18      0          False             True            0.112121   \n",
       "448      2372      0          False             True            0.200000   \n",
       "449         3      0           True             True            0.000000   \n",
       "\n",
       "     sentiment subjectivity color  \n",
       "0                       0.2   red  \n",
       "1                       0.2  gray  \n",
       "2                       0.2  blue  \n",
       "3                       0.6  gray  \n",
       "4                       0.0   red  \n",
       "..                      ...   ...  \n",
       "445                     0.2  blue  \n",
       "446                     0.2  blue  \n",
       "447                     0.6  blue  \n",
       "448                     0.2  blue  \n",
       "449                     0.0   red  \n",
       "\n",
       "[450 rows x 9 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "color = []\n",
    "\n",
    "for index, row in df.iterrows():\n",
    "    if (row['follows trump']):\n",
    "        color.append('red')\n",
    "    elif (row['follows clinton']):\n",
    "        color.append('blue')\n",
    "    elif (row['follows trump'] and row['follows clinton']):\n",
    "        color.append('purple')\n",
    "    else: \n",
    "        color.append('gray')\n",
    "\n",
    "df['color'] = color\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualization\n",
    "Plotting the tweets by sentiment analysis, there seems to be a greater variance in sentiment polarity with more subjective tweets. It is difficult to draw any further conclusions based on the the political leaning classificiation of users at this point."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZAAAAEGCAYAAABLgMOSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO3dd3zU9f3A8dc7l8si7L1BiFoQBY3g3uCqotaBVMX9s3VUrau1VWutdXRYR62oOFBBakWpe4+KqEERBEUQCFtCmNmXu/fvj883cFnkcrnL5ZL38/G4R/L93vd73/c349732aKqGGOMMY2VkugAjDHGJCdLIMYYY6JiCcQYY0xULIEYY4yJiiUQY4wxUUlNdADNqVu3bjpo0KBEh2GMMUll7ty5G1W1e839bSqBDBo0iLy8vESHYYwxSUVE8uvab1VYxhhjomIJxBhjTFQsgRhjjImKJRBjjDFRsQRijDEmKpZAjDHGRCWhCUREpojIBhH5pp7nRUTuF5GlIjJfRPYNe26SiCzxHpOaL2pjjIlMKBTim2++4ZNPPiE/v86esEkt0eNAngQeBJ6u5/njgRzvMQZ4GBgjIl2AW4FcQIG5IjJLVTfHPWJjjImAqvLss8+yatUqgsEgPp+Po446igMOOCDRocVMQksgqvoRsGkXh4wHnlZnDtBJRHoDxwJvq+omL2m8DRwX/4iNMSYyy5cvZ/Xq1QQCAUKhEIFAgLfffptgMJjo0GKmpbeB9AVWhW2v9vbVt78WEblURPJEJK+goCBugRpjTLjS0tI691dUVDRzJPHT0hNIk6nqZFXNVdXc7t1rTeVijDFx0a9fP8JXfBURunTpQkZGRgKjiq2WnkDWAP3Dtvt5++rbb4wxLULHjh2ZOHEi7du3x+fz0adPH84991xEJNGhxUyiG9EbMgu4QkSm4xrRt6rqOhF5E7hTRDp7x40DfpOoII0xpi6DBg3i2muvTXQYcZPQBCIi04AjgG4ishrXs8oPoKr/Al4DTgCWAiXABd5zm0Tkj8AX3kvdrqq7aow3xhgTYwlNIKp6dgPPK3B5Pc9NAabEIy5jjDENa+ltIMYYY1ooSyDGGNMEM2bACSfAWWfBN3XOqdF6tfRGdGOMabEefRSuvhpKSkAEXn0V5s6FPfZIdGTNw0ogxrRVb74JkyfD6tWJjiRp/fnPLnkAqLrvH3sssTE1JyuBGNPWhEIwbBgsXuy2RWDaNFcHYxql5qwkqu7H21ZYCcSYtua223YmD3Dveueck7BwktlVV0FW1s7trCyY1IbmBrcSiDFtzVdf1d5XWdn8cbQC114L7drBk09CdjbccQfsvXeio2o+VgIxpq2pazpxv7/542gFROCyy2DOHHjnnbp/tK2ZJRBj2pqbb4Z99925LQIvvJC4eEzSsiosY9qiuXPhiy8gPx+OOgq6dEl0RCYJWQIxpq3af3/3MCZKVoVljEkK//jHSq666kFuuuluzj9/Ot9/X5bokFqsq692NZNVj3jVUFoCMca0eG++uZkNG56ha9dCMjPL6N9/Kffc83yiw2qR3ngD/vGP6vvOOCM+He0sgRhjWrxXXllRbTs1NUjfvvls39561hePlTvvrHv/Sy/F/lqWQIwxLV5WVjqq1VfyC4VSyMqyt7CaevSoe/9uu8X+WvbTN8bEXDCohELa8IERuvHG3dm6tROBQCqqUFHhZ/36o/H5Ws/ysLHy5JO192VlVe+5HSuJXpHwOOAfgA94TFXvqvH834Ejvc0soIeqdvKeCwILvOdWqurJzRO1MaY+69aVcvfdM+jYMZ/KylS6dTuWa67Zr8mv26VLKrfeejG///2XlJZuJydnEI8+OjQGEbc+2dlQUOA62BUUuGnPZs+Oz7XELfrX/ETEB3wPjAVW45anPVtVF9Vz/JXAKFW90NsuUtXsxlwzNzdX8/Lymha4MaZeV1/9LO3bLyc11bVNBAJ+Ro/+OSefPDDBkZmmEJG5qppbc38iq7BGA0tVdZmqVgDTgfG7OP5sYFqzRGaMiUq7dit3JA8Any/AV1+tSFxAJq4SmUD6AqvCtld7+2oRkYHAYOC9sN0ZIpInInNE5JT6LiIil3rH5RUUFMQibmNMPSoqMqttB4OpdOjQLkHRmHhLlkb0CcALqhreZ2+gV6SaCNwnIkPqOlFVJ6tqrqrmdu/evTliNSYpFBXB2rWxXb9izJiTCQT8VFT4qahIo7i4K5dcMjJ2FzAtSiITyBqgf9h2P29fXSZQo/pKVdd4X5cBHwCjYh+iMa3Tn/8MXbvC0KEwZAisWBGb1z399N0YP/5SunUbx6BBJ3HHHReTnW0zJrVWiWxET8U1oh+NSxxfABNVdWGN4/YE3gAGqxesiHQGSlS1XES6AZ8C4+trgK9ijejGwIcfwgkn7FyKNSUFRoyAefMSG5dpueprRE/YRwNVrRSRK4A3cd14p6jqQhG5HchT1VneoROA6Vo90/0EeEREQrhS1F0NJQ9jjJOXV31ai1AIFi6s/3hj6pPQsqWqvga8VmPfLTW2b6vjvNnAiLgGZ0wrNWiQWz+qomLnvl69EhaOSWLJ0ohujImRU0+FsWPdUqwdOriBZ9Osg7yJgrVuGdPGpKTAiy+6ZVg3bnQjlq0EYqJhCcSYNkgEDjww0VGYZGdVWMYYY6JiCcQYY0xUrArLmDZIVcnPz6e0tJS+ffvSoUOHRIcUVxXlQe46ZQa+H1fRddxhXHbXAYkOqVWwBGJMG6OqTJs2jfz8/B3bEydOZNCgQYkNLE4qyoPMzjmaG9fPRiWFlAUhbn//L9zy2VWJDi3pWRWWMW3MokWLWLFiBRUVFVRUVBAIBHjxxRcTHVbc/HnsFA5c/ynpgQAZFeWkVQa4cd71VJTbcrhNZQnEtCmrPlvLZ49/w5b8rYkOJWG2bt1KqMYMisXFxQmKJv7aFa6otc8fCPDtl5ubP5hWxhKIaTP+NPZ9dj+gM+Mu7s+AQSl8+I+2OflT3759SUnZ+a8vIvSobyHtVqByn4OQsJmQQsDm9p3Z58BuiQuqlbAEYtqEr6Z9x53vjKGMTLbRke20Z/zVgwlVxnAu8yQxcOBAjjjiCFJSUvD5fHTu3JkJEyYkOqy4uem5E7k/59dU+nwEfKlsy+7I02f/O9FhtQrWiG7ahMWfbsJHn2r7SslgS/5WugzpnKCoEueggw5i9OjRlJeXk5WVhYgkOqS4um7Rn1m59Ga+eDufky7ck2vSfYkOqVWwBGLahD0P7krwgep/7lmU0mlgxwRFlHipqamkpradt4ABQ7MZMHR4osNoVawKy7QJI8/ag1uO+5wMyujAVtqzjVkP5JOSav8CxkSr7Xz8MG3eja8fwblfrmfdos3kHNGXDv32SXRIxiS1hH78EpHjRGSxiCwVkZvqeP58ESkQkXne4+Kw5yaJyBLvMal5IzfJqs++vdjvnJ/QoV/LGHm9cGEpY8cuZ8yY5dx++7pEh2NMoySsBCIiPuAhYCywGvhCRGbVsbLg86p6RY1zuwC3ArmAAnO9c61jt0ka339fxgEHVFJS0p9QKJUvvwyweHE+zz47sFmuHwrB8uVucakBA5rlkqaVSWQJZDSwVFWXqWoFMB0YH+G5xwJvq+omL2m8DRwXpziNiYsbblhPWVkGoZD7HFdZ6WfmzO7Ncu0NG6BbNxg6FAYOdI/wFQqNiUQiE0hfYFXY9mpvX00/E5H5IvKCiPRv5LmIyKUikicieQUFBbGI25iYKCmBYLD6v2BlZfNUChx5JGwOK6+vXOlWKjSmMVp6F5T/AoNUdW9cKeOpxr6Aqk5W1VxVze3evXk+3RkTiYkT2+Pz7ZyPKTU1wJAh65vl2j/8UHtfXl6zXNq0IolMIGuA/mHb/bx9O6hqoaqWe5uPAftFeq4xLd3553flhhs20rnzZrKyitlzz7V8/HGfhk+Mgc51jJ3s169ZLm1aEdGwOWKa9cIiqcD3wNG4N/8vgImqujDsmN6qus77/lTgRlU9wGtEnwvs6x36JbCfqm7a1TVzc3M1zz5mGcNnn8FBB7mGdHAN6UuXtt3G9MLCQpYtW4bf72fYsGGkpaUlOqQWRUTmqmpuzf0J64WlqpUicgXwJuADpqjqQhG5HchT1VnAVSJyMlAJbALO987dJCJ/xCUdgNsbSh7GmJ3GjIH8fHj8cUhLg1/8Ajp1SnRUiZGfn8+zzz6LqiIifPTRR/zf//0f6enpiQ6txWuwBCIiI1R1QTPFE1dWAjHG1PTQQw+xcePGHds+n48jjzySgw8+OIFRtSz1lUAiaQP5p4h8LiK/FJG2O3GQMaZVKi0trbYdDAYpKipKUDTJpcEEoqqHAj/HNVrPFZHnRGRs3CMzbdPs2XD44bDvvnDffZCgNjrTurz3HvTsCZmZkJsL27btfG633XbD59s5O6/f72e33XZLQJTJJ6I2EFVdIiK/A/KA+4FR4uZ//q2qtt61ME3z+vprGDvWDZAA+P57KC6Gm29ObFwmqS1eDMccs/OzyNy58JOfwBqv3+ZPf/pTysvLWbp0KT6fj6OPPpqcnJzEBZxEGkwgIrI3cAFwIm4sxkmq+qWI9AE+BSyBmNh45pmdyQNc8nj4YUsgpknuv792QXbtWigqguxsSEtL4+yzz97RiG4iF0kJ5AHcGIzfquqOykJVXeuVSoyJDb8fUlJ29i0FaEPrVcRCefnOH6Nx6uuRW/NPy5JH40XyZzZTVaeGJw8R+RWAqk6NW2Sm7bnoImjXDqr+kbOy4Hf2GaVeqq4+5u232fJDIYcf7n58mZlwzz2JDq7luP568NVYgHD4cMjISEw8rUkkCeS8OvadH+M4jIEhQ+Dzz+Hcc2H8eHj6abj44obPa4tCITjzTDjsMDjjDPx7DqFidh7BoJsU8Q9/gNdeCzt+yxaYOhWefBJ+/DFRUSdEnz6wYAHstRf06AGnnQbz5iU6qtah3voBETkbmAgMFpFZYU+1xw3qMyb29twTnmr0lGcJU1RURGFhIZ06daJjx2bs5f6f/8Drr+9oM2oHTGUCOSwF3O4PP4QTTgDWr4dRo2D7dneu3+8SdRtqKP7JT1wSMbG1qwrm2cA6oBvw17D924H58QzKmGSwcOFCXnrpJXw+H8FgkHHjxrH//vs3z8WXLXMNHmH6sXrH95mZYXNb3XYbbNwIlZVuOyUFrr4aXn21eWI1rVa9CURV84F84MDmC8e0ChUVbm6MdetcF5ihQxMdUcyVlZXx0ksvUVlZSaX3xvzWW2+x++67N09JZORI1zrsXTskKXzHMLLbuaeHDAmr/Vu1amfyAFf9tcbmHjVNV28biIj8z/u6XUS2hT22i8i2+s4zbdymTZCeDlOmuCqWnBz3CbiV2bZtGyk1ujr5fD42b26mRTGPPRauucYlkXbtSOnfj+4fvsCDD7qmo88+c6UQAI4/3nVIqJKZ6c43ponqTSCqeoj3tb2qdgh7tFfVlrGgtImfe++t6iQPZ58NZWWRnde/f+19f/hDbGNrATp27EjNeeSCwSBdu3Zt8mtPneomNvT7Ydw41/5dpzvucKW8+fPhhx/oe+huTJrkFoaqNg/g5ZfDJZe4F0xNda3Id9wRdXyLFi3i7rvv5vbbb+eJJ56guLg46tcyyW2XvbBExCci3zVXMKaFePFFV2ooLoZAAF5+2X3ajUT4QMBWLD09nTPOOAO/309aWhqpqamcdNJJtG/fvkmvO2cOXHYZbN3qap0+/BAmTvSefPpp2HtvN81LVRerLl1gt912PV5GxE0LU1bmHs8845JJFNavX8/MmTMpKytDVVm9ejUzZsyI6rVM8tvlKC1VDYrIYhEZoKormysok2CvvFI9EZSWuuqoSGRkRF5aSXI5OTn8+te/ZsuWLXTo0IHMHXVG0Xv//eprk1dUuCTCgw/ClVfufOLEE12ib8w6tDEYXZifn19tOxQKsWrVKhvF3UZF8hfVGVgoIu+KyKyqR7wDMwnUp0/tT6iRLge8fHntfb/9bdNjaqHS09Pp2bNnTJIHQLdutQsTfj91T+dy7bUxuWZjZGVl1UoU6enpljzaqEgSyO+BnwK347rzVj2aTESO80o4S0Xkpjqev1ZEFonIfC+BDQx7Ligi87yHJbRYuuYaN3VpVpYrUbRrBw89FNm5vXq57qVnngmHHupGbP3pT3EJc03eGuZOXUSwItjwwS1YZWUlJSUlqCoTJ9YuwG3dCt+U1TE7bAJKesOGDaNHjx74/X58Ph+pqamceOKJzR6HaRkanGhIVT+Mx4VFxAc8BIwFVgNfiMgsVV0UdthXQK6qlojIL4B7gLO850pVdWQ8YmvzunaFb75xg9VKS10vnkint1aFG25w1SspKfB//+eqv+pahDtKwYog/TIKWK99gD7IeSHmPPENo8/fK2bXaC4ff/wxH3zwASJC586d6dXrAiCr1nF/6f03nsw/qvrOCy5oniDD+Hw+LrjgAhYtWkRJSQkDBgygd+/ezR6HaRkaLIGIyAEi8oWIFIlIhffJPxbdeEcDS1V1mapWANOB8eEHqOr7qlpVGT8H6IdpHh07woUXuh48jVkb4bnn4LHHXAtwRQV89RVcemlMQzul/xes156AAIKSwqEXJN9Ykx9++IGPP/6YUChEMBiksLCQefPeqvPYjicfDief7Oq3/H7Xq+rOO5s5Ysfn8zFixAjGjBljyaONi2Sq0weBCcC/gVzc3Fi7x+DafYFVYdurgTG7OP4iILwlN0NE8nDrpd+lqi/FICbTVB995HpvVamocItExdBnG4bgkkcVoYLkW796zZo1OwYhAqgqmZmL6NDhlGoLHqWkwJ/uTIHslxMQpTH1i6hbhqouBXyqGlTVJ4Dj4htWdSJyDi553Ru2e6C3Ru9E4D4RGVLPuZeKSJ6I5BUUFDRDtG3ckCHVpzkVgQEDYnqJPhmFQPgYDEVIvpULO3bsSGqNFvPs7GwKCuDgg6FDB9hjDzdoPDs7QUEaswuRJJASEUkD5onIPSJyTYTnNWQNbpncKv28fdWIyDHAzcDJqrpj8h9VXeN9XQZ8AIyq6yKqOllVc1U1t3ukPYkMM6evonfvfHr1WsUhI/MiP/HKK93Mde3bu3fATp3g8ccjPz8YdOMU/vpXN6q9Dm/P7UoKIVwScYnjruM/iPwacVZZXs53b73FkvfeIxQ+hUgNI0aMoF+/fqSlpe14nHrqqaSlwf/+5xrPv/vO9UtoTRZNm8bsW29lw6JFDR/cCixbVsR7762jsLC84YOTjNQcTVvrANfzaQPgB64BOgL/9Eol0V9YJBX4Hjgalzi+ACaq6sKwY0YBLwDHqeqSsP2dgRJVLReRbriVEcfXaICvJTc3V/PyGvFm2EYdf+iHvPG/QwmvJurZcznr10fYFhIIuAENJSVwyCGub2okSktdA35p6c59L7/s6v5rKFpfxDWHf8nGbWlc89t0Druyzs8PzW7LypU89NhjVHpjLtIrK7nq2mvJ6tKlzuNVlRUrVlBWVka/fv2aPBCxpZt52mksHD4cX2UlKaEQh2ZkcFArnKmgyp13zqa09D2CQR8gHHTQRH7609iWyJuDiMz1anyq728ogcSTiJwA3Af4gCmq+icRuR3IU9VZIvIOMAI3KzDASlU9WUQOAh4BQrjS0H2q2uDHXEsgkcnIKKG8PJOdCcR90leN8zJ3Bx9cu73E56s+EWAL98DNN7PJ79+5KJYq/VS5qBW/SUbqgyuuYHbHjgTClgjssGUL1/z97wmMKn4++mg9b731OH7/zr/fsrIM7rjjBny+5Bo3U18C2dV6IAug/oplVd27qUGp6mvAazX23RL2/TH1nDcbl1hMHFRUpFOzkbpZfFfHrDnB5BrjsS0lZWfyABChMIkSYDyVrF+PdupUbd+2jh0p27aNjA6tb3q9777bSChU/UNXamqAdetK6devdlftZLSrXlg/bbYoTIvi95dTUVGzBNIMhg51Cx2Fq7kWaQvXXpXNqtVKIJ1tgXIAMnv0QGrUeLTfvr1VJg+A3XfvysqVNSfcTKV379jMWtAS7Go23vyqB1CG+8Q/AjeAL7++80zy271/VVPSzkbq3r3qmKIk1j74wM3+G27atPhfN4bOufBCfKGQG1Cpij8Y5OdXXZXosGIqFHL9Im65pXFLwx75z3+y53ffkRoIkF5WRnpZGaNr/r5bkSOO6I3ffwiVlamUl6dTUZHG6NETkq76alciaUQ/E9d99gPcR9JDgetV9YW4Rxdj1gYSmYsugilTyujVawOqKWzeHKRPn4F1TnMVc8EgTJ7slmG99FLo27cZLhpbFcXFLH73XVJSU9nj6KNJTU++MSr1CYVg4EBYvXPxQ+6/v/o8jw2ZN3ky25csYehZZ9E7t1a1equzaNFWVq7czj77dE3a0kfUjegi8jUwVlU3eNvdgXdUdZ+4RBpHlkAic9NNrhdteNX9yJFuUHmLccstbiR2KATDhrmpV0zc3Xxz7QHwKSlJ11RlGqm+BBJJ5WxKVfLwFEZ4nklS117retNmZLiZM7Ky3KfMFuOvf4U//tG9a6nCwoV1L2RlYq6ufg6hkHuYtieSRPCGiLwpIueLyPnAq9ToOWValx493Hvy3Xe7heu++MJNrNti1NUlNrxOpYUoLW25H8unT99E+/ZFpKWVM2LEjxGfd0wd/SLT02Oy1IhJQhGNAxGR04BDvM2PVXVmXKOKk2SswiorK2Px4sUEg0GGDh1Kh0b0WCkvL2fx4sVUVlY2+twWrX17KCqqvT/CMU1lZfDSS7BtGxx9tJt9JZbeemsN77wznaysIkpKsjnmmAmMG9dy2nLeeGMrxx9f9bcggNKt21YKCjrt6rQdTjxx54KIPh+89RYcddSuzzHNTNXNgp2f71awHLOraQYb1uhxIDXMBoK4gXtfNCkSE7GSkhIeeeQRSr2R2SkpKVx44YX06NGjwXNLS0t55JFHKPFWFhQRLrjgAnrFaF6M4mL3yXNXK6nGzZVXwp//XH1fPSO9ayopgf33h5UrXbWLiHszPOyw2IRWWFjOBx9MpV07N21Fu3ZFvP/+VPbb7xq6dm0ZjemXXlrVuCU7vm7c2DHi8199FTZscD/Dvfeu3XHOJJgqnHOOm8Wh6o/8zjvhV7+K+aUimc79YuBz4FTgdGCOiFwY80hMLR9//DFFRUUEAgECgQDl5eW89lpktYeffPIJ27dv33FuRUVFxOeGq6io4L///S8PPfQQjz76KA8++C9+/et/MnbsHLKylHvuafRLNt2dd1ZfC6NPH/gxsmqYRx+FZctcAaakxCXCSy6JXWhffrmx1j4R+OqrwthdpIkqKppe39SjB+TmWvJokT77zCWP4mI3LVBJiVujJ3yZ6hiJ5PPj9cAoVS0EEJGuuBLJlJhHY6rZtm0boRqtk0V1Vd3E+Nxw06dPZ9WqVdWmHe/QAY488j1EgvzhDwczahSMHdvol26aKVPco5HWr6+9kF8sJ2nu2zcbn6/6yHOfL0jfvu1id5EmuvTSCv74x/A9it8fACwbtAobNtQegOvzwebNrkdMDEXyUaQQ2B62vd3bZ+IsJycHf9ja5KmpqQwdGtnCSUOHDq117pBGVvaXlZWRn59fLXlUSUsLkJubR2mpmzk2WRx5ZPX/ofR0OOKI2L3+sGEdgQOoqPDveIgcyE9+EnkVUbzdfntPfvaz9Yi4GY0zMspYtMi6UbUa++1XvV+1iJvQNA7TOkdSAlkKfCYiL+OGJY8H5ovItQCq+reYR2UA2GeffSgsLOTTTz9FVdljjz04pq5uMHUYMWIEhYWFfPLJJ6gqOTk5HHvssY26fkoDXWuCQR+ZmdAvidaJHDcO7roLbrzRrXV12GHwxBOxvcYf/3gMr7yyO0uXFjB0aPcWOfvqCy+ErySYnIPbTD369oVZs+Css6CwEHJy4JVX4jItUCQDCW/d1fOqmjTTjCZjLyxwU36DawhvznMBXnrpJRYtWkQgEPBez32gCQT8vPHGSaSmjuCjj6KsC7//fvjtb907+UknwdSpMS9i10fVtS8m2VRbxjROMBiTP/IWOZ17c0vWBBI3qm7Rpk6d6v0jU1U+//xzVqxYQUZGBsFgkO3bQ1RU7EOfPjkce2yUPbFeew3OOGNnw15GBkyYEPviQBgNhShbv560Ll3wha+aGC9FRXD66a5O+vLL3RwxJq4Kpk8nc6+9yN5rr0SH0qo0tRuvaW0WLHD1OZs2ueQxdSr87Ge1DhMRxowZw5gm9iOv5fXXq/cKKSuDN9+M7TXCbPrqK6Y++yzbM111zYk9ejCqMRM4NdaWLdC5887tiy929/xC0k0hlxTW/+tfdL3ySrp57XUFXbvSfWPtHnEmtmz8aFsUDLpuU+vXu+qj0lI47zyaZ7ZET69eteu9Il25MArPPvMMW9q1I5iaSjA1ldd+/JH1H30Ut+tx8MG19/3nP/G7XhvX8eqrSa2sRHCjW7oVFrJi+PBEh9XqRTIOpNZ/Ql37oiEix4nIYhFZKiI31fF8uog87z3/mYgMCnvuN97+xSLSuNbhKDy++yRezz6eN9uN48nB5zbq3B9//JGXX36ZF198kRUrVjTuwvfdBx07Qrt2cNlljTu3Phs2uAW3w6WmNm5u7l1YcM3jbE3pRLFk8fU+9fysLr+cQPc+lKdmUZ6SQWV6O/jnP2Ny/Zoqi4vZnJ1dbb4NUWVtjO63ThGOS5n7yjpO7fohR2XN5vbjZzd8gqlTRnl5rSXQ2q9bV9/hrV55eTnvvPMOM2bMYM6cObW69MdKJFVYDwD7RrCvUUTEBzwEjAVWA1+IyKwa65pfBGxW1aEiMgG4GzhLRIYBE4DhQB/gHRHZXVXjMvnQlJzzmLD0P2RSQgpQsiKTpwYpk1Y80+C5P/74I48//viORuhvv/2WM888k5ycnIYv/MADcM01O7cfecT15X7++SjvxNOlS+1pPyorYzIh4Te/eYa97rsYcP/Ee89/hkX91jNs9dvVjlu+uROHbJvPcZX/JotiPpTjuG5ZDucdUseLNpEvMxN/IEBF+LTqqnQMr2KKtUMPdfOl7MLCDwoYd1IaWzmIIH4+e6OYH/f6gIe+OSJ+cbVSwZQUUsPeJBUob9dyxuGdDAkAACAASURBVN40p2AwyJQpUygsLCQYDLJ06VLWrl3LaaedFvNr1VsCEZEDReTXQHcRuTbscRtuDfOmGg0sVdVlqloBTMd1EQ43HnjK+/4F4Ghx3YnGA9NVtVxVl+O6Go+OQUx16r927Y7kAZBFKcPXfBvRuZ9++umO5AFQWVnJhx9+GNmFf/e72vv+/e/Izt2V9HQ3JDsz080r1a4dTJrkhhY3Uc+/XA+ET5IBe655p9Zxjz4K64vbM4ULeZArWVCWU+ccibEgqvzsxRfxV1SQVlaGv6KC3ZcuZbd4zsMyc6Ybrh3u2WerbT585SKKaUcQN16nhHY8tzDpVkloEVYde+yO5c8UCKSmwr/f5Ouvoa01heTn57NlyxaC3liQQCDAwoULd0yJFEu7+g9KA7K9Y9qH7d+Gm9KkqfoCq8K2VwM1W2p3HKOqlSKyFejq7Z9T49w6Z6sTkUuBSwEGDIiuP35qqLJWpvVroM5ja6prEF5d++oUz0UWzj0XRo921VYDBsCBB8bkZX0aqLWCel0diMvLa08BXlERkxBqC4XYfelSfvHww6zp04f2RUUMKChATjopThf0/Pije/f67js4pHbRKhBQQjX+siqtX0tUBr/2GmvuuYfKKVMIZmWx+Ia3OHNsN1JS3N/VAw+4fgxtQTAYrNVtX0TiUo21qyVtP/TGeBygqn8Ie/xNVZfEPJI4UdXJqpqrqrndu3eP6jVW9uhHSdhgq2KyWNAtsm6C++67L6lhn3T9fj/7779/ZBc+//za+/aJ4SfUPfZwg41ilDwA8k+8vNoK6gqsyaw9en7ixOpDPrKy3AKEceH3w0kn0bmsjL0WLmTgypWI3+96ocVbt251Jg+AM68fiJ+dWTOTYo7p8mX8Y2ql+t5wAwO/+44+s7/kzEu6UVTkZlwuK4OrrnJzoLUFAwYMIDU1dUcS8fl89OnTh6w4jLGKpBdWuohMFpG3ROS9qkcMrr0GCK907+ftq/MYEUkFOuKmUYnk3JiZlP8ML/Q9jW98w/jWtycze4znnPXPRXTubrvtxumnn06vXr3o0aMH48aNY999I2w+evBBOPtsN3JPBEaNghY+jmXUy3/g69GXECSFEMLKrD3os7X2KkT77ecGx44e7RYUvOUWt9pd3Eyb5sZh7LGHm8/k00+hZ884XrBhR180mCm3rGRkynyG8AOn9fucafkx6Z/Spq1dW7uJLy0NFi9OTDzNLT09nYsvvpghQ4bQtWtX9tprL37+859HPZh4VyJd0vZfwFzclO4AqOrcJl3YJYTvgaNxb/5fABNVdWHYMZcDI1T1Mq8R/TRVPVNEhgPP4do9+gDvAjkNNaLbQMJWZvt2N5akRw+XYI3B/Ul07159mFFmJnz9tZvVwzReU5a0rVTVh1X1c1WdW/VoakCqWglcAbwJfAvMUNWFInK7iJzsHfY40FVElgLXAjd55y4EZgCLgDeAy+PVA8u0QKqud1rXrjBwoFuwPZZT6pqklpXlCpxZWa4HfGamWwHAkkfsRVICuQ3YAMwEyqv2q+qmuEYWBy2+BLJhg/vLDwTglFMgwpl325wZM9x6IFUfMf1+t7Tg668nNi4TV8HSUlbffTehdevIGjeOnnXMnBBuwwb4/nvXRyTK/jPG05SpTCZ5X68P26fAbrEIzHjWrHGfpLdvd92TbrsNPvggJl1rW51PP61ePxEItPi2oabaWBDg4AOWERJlv73SmP5S2/r3C5aVUTh8OL1Xr8ZXWUnoiSdY8etfM6jmypRhevSo3ZPaxFaDVViqOriOR9v6620Of/6zGyRYXu7eEIuLqw8iNDsNHerqJaqINP0j5ltvwfDhbjDlNde430ELsbEgwL77rWX5qsGsWj2Il97ozR5D2kiLsGfd5Ml0XLOGtEAAnyr+QIC+f/kLGqcR1iYykUxlkiUivxORyd52joj8NP6htTEbNtQe91Fo63bV6ZJLXI+07Gy3PGKnTvDUUw2fV58vv4RTT4VFi2D1apg8uUUl74MOWsb69b0JBNIoL8+gvDyDDZva1kfr4MaNSI3qdl8wSChug4dMJCJpRH8CqAAO8rbXAHfELaK26rTTag+MOPXUxMXTkqWlwUcfuX7Azz0HS5ZAJNN3L1jghsD/97/VRzHOnOkmlKxSUgLTp8c+7igpKVRWhk/+IGzf3r7e41ujDj+t/pm1MiWFgsGDm2daflOvSNpAhqjqWSJyNoCqlkg8OhS3dRMmuE+/f/qTm5fqvPOI29werYHPB4cfHvnx06btHIqckuIG9736qvs+O9tNJhlebdWC3piEEKmpQQIBl0REQnTqtAWI3+zFLU3n0aNZ9/jjZF95JRnFxWzMyaHjG28kOqw2L5JeWLNxYzU+UdV9RWQIME1V4zb3VLy0+F5YJj5UXZIIb3jPznaljBNPdNWHI0a4tVEqK13p75FH4JxzEhdzmI0FAUaPyWflykGkpARJS6tgYO+1LFyyR6JDM21EU3ph3Yoba9FfRJ4FDgbOj214xsRRRYWbzyJcKLRzyvUePdwos4cecknktNNct+AWolt3P8uWDeWwMUsoLAxx02+7c+6FljxM4kW0pK2IdAUOwM2LN0dVk3J+SyuBtGEDB8LKlTu3U1Jg/nzX88oYs0tNGYkObqZbH26G3sNEJPYTyxsTL1Vrv4dLS3Njb4wxUWuwCktEpgB7AwuBqq4rCrwYx7iMiZ2KiurtH+Aa4Vetqvt4Y0xEImkDOUBVh8U9EtNmlJUFeemlzRx9dAe6d09r+ISmSk+HIUNg6dKd07SGQhDptPotWOWWIkqWrKLD/j9JdCimDYqkCutTbwlZY5rsvPNWkZmZwtlnd6VHDz/9+0e2dniTvfaaawdJS3MJ5aGHYO+9m+facbKk96H4Oren/ehhlEs6BdPfTXRIpo2JpBvv4cAsYD1uMkUBVFWT7r/PGtETq6wsSGZm1WeWqqFEyrnnrubpp5u+HnuDVN10MR06uHEfSWzJadczdOZfwn6KUEIW7bQ4ovNXrVrFq6++SmlpKUOHDuX444+vtvCZMeGa0o33ceBcYAE720CMabSXXtqMW5G4+jjU995Lb54ARKBLl+a5VpylfPxRtW0Bsiip++AaCgsLmTp1KgFv4OT8+fMpLy/n9NNjsVK1aUsiSSAFqjor7pGYVu/oozvUub937/I695v6BXv0hhqd6YOkRPQPvWTJkmrrY1dWVvLdd7VXjTSmIZG0gXwlIs+JyNkiclrVoykXFZEuIvK2iCzxvnau45iRIvKpiCwUkfkiclbYc0+KyHIRmec9RjYlnoioukWVlyypPo+SiVj37mn067cBV+HiHj5fJR9/3CfBkSXWfTfM5foT32Lxgq0Rn7PbJ89QRPuwnySsOOXaiM71+/2kpFT/1/f5fPUcbUz9Ikkgmbi2j3HASd6jqbPx3gS8q6o5uOVob6rjmBLgPFUdDhwH3CcincKev15VR3qPeU2MZ9fKy2HcODdh38iRrvfO1sj/2c1Oq1b15NxzV9O3bwG5uWsoKkohI6NtvnmVlVbyz9RfcNm9B3PLaz+j3d7DueXnHzV8IpDaKZus7T+y5JhfsGTEz1j/8IsMnXlvROcOHz6czMzMHUnE7/dz5JFHRn0fpu2KaCR6zC8qshg4QlXXiUhv4ANV3eXcDN7a7Ker6hIReRJ4RVVfaMx1o25Ev/VWuPfenTO2pqfDxIkwZUrjX8sYz+X9HuPuNVeTjWv4DpLC1+zNvvpV3K9dWlrKnDlzKC4uJicnhz32sKlRTP0a3YguIjeo6j0i8gCuhFyNql7VhHh6quo67/v1QM9dHSwio3Gj4H8I2/0nEbkFrwSjqvGrSP/ii+rTfZeXw9wmLwtv2rjeG1eQyc6/Kx8h9qR5ForKzMy0Uodpsl21uX3rfY2q36uIvAP0quOpm8M3VFVFpN5ikFdCmQpMUtWqxoff4BJPGjAZuBG4vZ7zLwUuBRgQ7ap1e+0F77+/c0I+v9/mUDJNtqFjH0o3ZO4ogYSA5QzC/rJMsohkHMgZqvrvhvY16qIRVmGJSAfgA+DO+qqrROQI4DpVbbBdJuoqrOJiOOww+P571xW0Z0+YPRu6d2/8axnj2b41wKudf85J+ioB/ATxcefYZ/jrW8cnOrS4Wr9+PcXFxfTq1Yt27dolOpxW69tv3fyhw4dDv35Ne62mjAP5DVAzWdS1rzFmAZOAu7yvL9c8QETSgJnA0zWTh4j09pKPAKcA3zQhloa1aweffQZffeWWnR01yrWDGNME7Tv6OaX4OX536juUrN3Gib8ZzV/PHpTosOJGVXn55ZdZtGgRKSkpqCoTJ05k4MCBiQ6t1bnlFvjLX9zEC4EATJ3qVimItXpLICJyPHACcCbwfNhTHYBhTVlQypsefgYwAMgHzlTVTSKSC1ymqheLyDm45XQXhp16vqrOE5H3gO648VPzvHOKGrqujUQ3JnGWLl3KjBkzdgxgBGjXrh3XXXddAqNqfebPhwMOqN5sm5npJmGI9nNvNCWQtbj2j5OB8Bbj7cA10YXhqGohbpXDmvvzgIu9758Bnqnn/KOacn1jTPPbvHkzNT+wFhcXEwqFao1LMdFbvtw104YnEICCgqZXZdVUbwJR1a+Br0XkOVUN1HecMcZEolevXrha5506d+5sySPG9trLVVuFy8iAXnV1aWqiSH5zo73R4t+LyDJvBPiy2IdijGnN+vfvz+GHH47P58Pv95Odnc3EiRMTHVarM2QI/OtfLmm0awedOrnJqOMxV2YkvbC+w1VZzQWCVfu9aqikYm0gxiReWVkZpaWldOjQwaZQiaOiItiwAfr2bXqfn6b0wtqqqq837fLGGONkZGSQkZGR6DBavexs94inSBLI+yJyL24J2x2jvVX1y7hFZYwxpsWLJIGM8b6GF18UsJ5QxhjThjWYQFTVJswxxuwQCASYN28excXFDB48uFEDAYuKKpk8eR7btxcxcuQgxo8fFL9Ak1woFOLrr79m69at9O3bl5ycnESHVEuDCUREegJ3An1U9XhvffQDVfXxuEdnjGlRKisreeyxx9i0aROVlZXMnj2b448/nlGjRjV4bnFxJb///WO0a1eIz1fJF1/M5ocfjuXaa/drhsiTSygUYurUqaxZs4ZAIIDf7+fggw/m8MMPT3Ro1UTSjfdJ4E2gatWf74Gr4xWQMablWrRoEZs3b6ayshJwpZE33ngjonOfffY7srI24/dXkpICfn+ATZveJBRq/iUlWrr8/HzWrl27Y9R+IBDgo48+qjaKvyWIJIF0U9UZeOuhq2olYd15jTFtR1lZWa3R5IFAoNa+uhQVleG9jezg81USDFoCqamsrKzWoEsRScoEUuzNXaUAInIAYMvxGdMGDR48uNq2z+dj8ODBtd7s6nLYYdWPq6z0sXXrAPx+G4leU//+/aslZRGhS5cuZGZmJjCq2iL5zV2Lmz13iIh8AjwNXBnXqIwxLVL37t2ZMGECHTp0IC0tjSFDhnDGGWdEdG5ubldGjZpAUVFHKirS2L59MDfeeFacI05O2dnZnHfeeXTp0oW0tDT69+/PueeeG1Gibk4RLWkrIqnAHrjZbxcn69xYNhLdGGMar76R6A2WQETkDCBTVRfi1t54XkT2jUOMxhhjkkgkVVi/V9XtInIIbgr2x4GH4xuWMcaYli6SBFLV4+pE4FFVfRW3Frkxxpg2LJIEskZEHgHOAl4TkfQIz6uXiHTxpohf4n3tXM9xQRGZ5z1mhe0fLCKfichSEXneW/7WtAZffgknneTWoJ8yBSJoozPGVFe8YgUvXXcdj199Ne/87ndUFhfH5TqRJIIzcQMJj1XVLUAX4PomXvcm4F1VzQHe9bbrUqqqI73HyWH77wb+rqpDgc3ARU2Mx7QE337rEscrr8DHH8OVV8I//pHoqIxJKoGtW3n0wQdZkJnJ6s6d+Qx4/uab43KtBhOIqpao6ouqusTbXqeqbzXxuuOBp7zvn8I1zkdEXD+2o4AXojnftGBPPgklJTu3S0rgb39LWDjGJKP8N96gNC2NkLeCVKXfz/L27SlZtSrm10rUCJ6eqrrO+3490LOe4zJEJE9E5ohIVZLoCmzxRsQDrAb61nchEbnUe428goKCmARv4qSF9XE3LV8gAM8+C3/9K8yZk+hoWoZ6x4rE4f8rDoscOiLyDlDXKrzVylKqqiJSX0X3QFVdIyK7Ae+JyAIaOQpeVScDk8GNA2nMuaaZnX8+PPggVNXXZmXBddclNCTTclVWwlFHwVdfuUSSmgr33w8XtfEK7QHHHkvW559T6fMRSk0lNRBgcHExWf36xfxacSuBqOoxqrpXHY+XgR9FpDeA93VDPa+xxvu6DPgAGAUUAp28wY0A/YA18boP04z23BP+9z845RQ48kh46CG46qpER2VaqFdfhXnz3OeNigpX43nFFdbvwt+xI5f86lfsU1bGgM2bOUCEs+68My7XilsJpAGzgEnAXd7Xl2se4PXMKlHVchHpBhwM3OOVWN4HTgem13e+SVIjR8LMmYmOwiSBwsLayaKiwpVG0tp4v8ys/v05+d57436dRLWB3AWMFZElwDHeNiKSKyKPecf8BMgTka+B94G7VHWR99yNwLUishTXJmJrkxjTxhxySPUEkpoKo0ZZ8mhOEc2F1VrYXFjGtC5vvAGTJsHmzTB6NPznP9Czvi45Jmr1zYWVqCosY4xpsuOOgx9/THQUbZdNxG+MMSYqlkCMMcZExRKIMcaYqFgCMcYYExVLIMYYY6JiCcQYY0xULIEYY4yJiiUQY4wxUbEEYowxJiqWQIwxxkTFEogxxpioWAIxxhgTFUsgxhhjomIJxBhjTFQSkkBEpIuIvC0iS7yvnes45kgRmRf2KBORU7znnhSR5WHPjWz+uzDGmLYtUSWQm4B3VTUHeNfbrkZV31fVkao6EjgKKAHeCjvk+qrnVXVes0RtjDFmh0QlkPHAU973TwGnNHD86cDrqloS16iMMcZELFEJpKeqrvO+Xw80tAjlBGBajX1/EpH5IvJ3EUmv70QRuVRE8kQkr6CgoAkhG2OMCRe3BCIi74jIN3U8xocfp25R9noXZheR3sAI4M2w3b8B9gT2B7oAN9Z3vqpOVtVcVc3t3r17U27JGGNMmLitia6qx9T3nIj8KCK9VXWdlyA27OKlzgRmqmog7LWrSi/lIvIEcF1MgjbGmNZAFT7+GPLzYeRIGDEiLpdJVBXWLGCS9/0k4OVdHHs2NaqvvKSDiAiu/eSbOMRojDHJ6bLL4IQT4Je/hDFj4LHH4nKZRCWQu4CxIrIEOMbbRkRyRWTHnYrIIKA/8GGN858VkQXAAqAbcEczxGyMMS1fXh488wwUF0NREZSWwhVXuK8xFrcqrF1R1ULg6Dr25wEXh22vAPrWcdxR8YzPGGOS1tq1kFrjrT0lBTZtgr613k6bxEaiG2NMazJyJFRWVt/XsSP06hXzS1kCMcaY1mTAAJgxA7KzXUmkXz94+23w+WJ+qYRUYRljjImjE0+EbdtcO0h2dtwuYyUQY4xpjUTimjzAEogxxpgoWQIxxhgTFUsgxhhjomIJxBhjTFQsgRhjjImKJRBjjDFRsQRijDEmKpZAjDHGRMUSiDHGmKhYAjHGGBMVSyDGGGOiYgnEGGNMVBKSQETkDBFZKCIhEcndxXHHichiEVkqIjeF7R8sIp95+58XkbTmidwYY5LA8uVw7LGQkwOTJrmZeeMgUSWQb4DTgI/qO0BEfMBDwPHAMOBsERnmPX038HdVHQpsBi6Kb7jGGJMktm6FAw6Ad96BpUvh+efhuONANeaXSkgCUdVvVXVxA4eNBpaq6jJVrQCmA+NFRICjgBe8454CTolftMYYk0Q++QTKyiAUctvl5fDll1BQEPNLteQ2kL7AqrDt1d6+rsAWVa2ssb9OInKpiOSJSF5BHH6AxhjToqSn70weVUIhSIt9TX/cEoiIvCMi39TxGB+va9ZFVSeraq6q5nbv3r05L22MMc3v0ENht91cIgHIyoIJE6BTp5hfKm5L2qrqMU18iTVA/7Dtft6+QqCTiKR6pZCq/cYYY9LSXDXWPffA99/DIYfAL38Zl0u15DXRvwByRGQwLkFMACaqqorI+8DpuHaRScDLiQvTGGNamOxsuP32uF8mUd14TxWR1cCBwKsi8qa3v4+IvAbglS6uAN4EvgVmqOpC7yVuBK4VkaW4NpHHm/sejDGmrRONQ9eulio3N1fz8vISHYYxxiQVEZmrqrXG7LXkXljGGGNaMEsgxhhjomIJxBhjTFQsgRhjjImKJRBjjDFRaVO9sESkAMhv4st0AzbGIJxkYffbutn9tm6xut+BqlprKo82lUBiQUTy6urO1lrZ/bZudr+tW7zv16qwjDHGRMUSiDHGmKhYAmm8yYkOoJnZ/bZudr+tW1zv19pAjDHGRMVKIMYYY6JiCcQYY0xULIHUQ0SOE5HFIrJURG6q4/l0EXnee/4zERnU/FHGTgT3e62ILBKR+SLyrogMTEScsdLQ/YYd9zMRURFJ6q6fkdyviJzp/Y4XishzzR1jLEXw9zxARN4Xka+8v+kTEhFnrIjIFBHZICLf1PO8iMj93s9jvojsG5MLq6o9ajwAH/ADsBuQBnwNDKtxzC+Bf3nfTwCeT3Tccb7fI4Es7/tftPb79Y5rD3wEzAFyEx13nH+/OcBXQGdvu0ei447z/U4GfuF9PwxYkei4m3jPhwH7At/U8/wJwOuAAAcAn8XiulYCqdtoYKmqLlPVCtzKhzXXch8PPOV9/wJwtIhIM8YYSw3er6q+r6ol3uYc3FLCySqS3y/AH4G7gbLmDC4OIrnfS4CHVHUzgKpuaOYYYymS+1Wgg/d9R2BtM8YXc6r6EbBpF4eMB55WZw5uWfDeTb2uJZC69QVWhW2v9vbVeYy61RO34lZHTEaR3G+4i3CfZpJVg/frFfH7q+qrzRlYnETy+90d2F1EPhGROSJyXLNFF3uR3O9twDneyqivAVc2T2gJ09j/8Yi05DXRTQskIucAucDhiY4lXkQkBfgbcH6CQ2lOqbhqrCNwpcuPRGSEqm5JaFTxczbwpKr+VUQOBKaKyF6qGkp0YMnESiB1WwP0D9vu5+2r8xgRScUVgwubJbrYi+R+EZFjgJuBk1W1vJlii4eG7rc9sBfwgYiswNUZz0rihvRIfr+rgVmqGlDV5cD3uISSjCK534uAGQCq+imQgZt4sLWK6H+8sSyB1O0LIEdEBotIGq6RfFaNY2YBk7zvTwfeU6+1Kgk1eL8iMgp4BJc8krl+HBq4X1XdqqrdVHWQqg7CtfmcrKp5iQm3ySL5e34JV/pARLrhqrSWNWeQMRTJ/a4EjgYQkZ/gEkhBs0bZvGYB53m9sQ4Atqrquqa+qFVh1UFVK0XkCuBNXI+OKaq6UERuB/JUdRbwOK7YuxTXeDUhcRE3TYT3ey+QDfzb6yuwUlVPTljQTRDh/bYaEd7vm8A4EVkEBIHrVTUpS9QR3u+vgUdF5Bpcg/r5SfwBEBGZhvsA0M1r17kV8AOo6r9w7TwnAEuBEuCCmFw3iX9mxhhjEsiqsIwxxkTFEogxxpioWAIxxhgTFUsgxhhjomIJxBhjTFQsgZg2R0RGhs++KiIn72pG3hhd8wgROShGr7XCG6tRc/9lInJeFK/XSUR+GbbdR0ReaOCcHT8zETlFRIY19rom+Vk3XtPmiMj5uNl1r2jGa94GFKnqX2LwWitw8W9s6mt5rzcIeEVV94ry/Ce983eZdEzrYyUQkzREpJ2IvCoiX4vINyJylrd/PxH5UETmisibVbOMisgHInK3iHwuIt+LyKHeyOTbgbNEZJ6InCUi54vIg945T4rIw96Egsu8ksMUEfnWe6OsimWciHwqIl+KyL9FJNvbv0JE/uDtXyAie3pv0JcB13jXPLTGfR3u7Z8nbn2K9t51Xwk75kEv8VW5wXv9z0VkqHfMbSJynff9EBF5w/uZfCwie3r7e4rITO9n+LVXKroLGOJd/14RGSTeuhLez2F4WBwfiEhu1c/MO/9k4F7v/CEi8mXY8Tnh26Z1sQRikslxwFpV3cf7tPyGiPiBB4DTVXU/YArwp7BzUlV1NHA1cKs3vfctuPVMRqrq83VcpzNwIHANbgqIvwPDgRFe9Vc34HfAMaq6L5AHXBt2/kZv/8PAdaq6AvgX8Hfvmh/XuN51wOWqOhI4FCiN4GexVVVHAA8C99Xx/GTgSu9nch3wT2///cCHqroPbv2IhcBNwA9ebNfXeJ3ngTMBvMTcO3xKF1Wd7f2MrvfO/wHYKiIjvUMuAJ6I4H5MErIEYpLJAmCsV6o4VFW3AnvgJj58W0Tm4d7Yw9cqedH7OhcYFOF1/utNa7EA+FFVF3iztC70XuMA3CJEn3jXnASEr9DY2Gt+AvxNRK4COnnLAzRkWtjXA8Of8EpDB+GmnZmHm8Osau2Ho3CJDVUNej/DXZmBm+sNXCKJpJrqMeACEfEBZwFJvbqhqZ/NhWWShqp+L26djhOAO0TkXWAmsFBVD6zntKpZg4NE/vdedU4o7Puq7VTvtd5W1bNjcU1VvUtEXsXd1ycicixQSfUPeBk1T6vne7zztnglmiZR1TUiUigie+OSwWURnPYf3FxM7wFzk3VOLdMwK4GYpCEifYASVX0GN7njvsBioLu4NR0QEX94nX09tuOmbI/WHODgsLaHdiKye7TXFJEhXinnbtxMsnsC+cAwEUkXkU54M8eGOSvs66fhT6jqNmC5iJzhvb6IyD7e0+/iliRGRHwi0nFXsXmeB24AOqrq/IbuTVXLcBMZPoxVX7VqlkBMMhkBfO5Vy9wK3OG1aZwO3C0iXwPzcNU3u/I+7s15XlVDfGOoagFusalpIjIf9wa+ZwOn/Rc4ta5GdOBqr1PAfCAAvK6qq3DVR994X7+qcU5njKisqQAAAKZJREFU7/hf4dpqdoTnff05cJH3M1nIziVdfwUcKSILcFVsw7wSwideDPfWEfsLuNmmZ9Rzb9OB670OAEO8fc/iSmxv1XOOaQWsG68xrYSIPAB8qaoJ/9Tv9QbrqKq/T3QsJn6sDcSYVkBE/giMwa31nVAiMhMYgmuwN62YlUCMMcZExdpAjDHGRMUSiDHGmKhYAjHGGBMVSyDGGGOiYgnEGGNMVP4f7H/2bc9H58IAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.plot(kind='scatter',x='sentiment subjectivity',y='sentiment polarity',color=color)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Challenges\n",
    "- **Data quality:** duplicate tweets are being overlapped\n",
    "- **Sentiment analysis scores:** tweets may also overlap due to having identical analysis scores\n",
    "\n",
    "### Potential Solutions\n",
    "- **Radius:** increasing the radius of points were multiple tweets overlap \n",
    "- **Small multiples:** separating tweets by political leaning may help declutter points and reduce overlap\n",
    "- **3D plane:** if applicable, a third plane could be introduced to help distiguish points\n",
    "\n",
    "---\n",
    "\n",
    "## Next Steps\n",
    "- **Data**\n",
    " - Increase sample size of data\n",
    " - Improve quality of data (duplicates)\n",
    " - Develop method for identifying political leaning\n",
    "\n",
    "- **Sentiment Analysis**\n",
    " - Research additional models that may better suit social media data\n",
    " - Assess perceived accuracy of sentiment scores \n",
    "\n",
    "- **Visualization**\n",
    " - Address the overlap of data points\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}