13. JSON City Similarity.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "43091061",
   "metadata": {},
   "source": [
    "# Goal of Analysis:\n",
    "\n",
    "## 1. find the next city the tourists will most likely visit based on their last city choice\n",
    "\n",
    "## 2. cluster cities into two groups: high intent and low intent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "db11a8f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1915da32",
   "metadata": {},
   "source": [
    "### Part 1: Load Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "0b167950",
   "metadata": {},
   "outputs": [],
   "source": [
    "def clean_json(d):\n",
    "    \"\"\"\n",
    "    clean the json data, make the data easier to be processed\n",
    "    \"\"\"\n",
    "    assert len(d['cities']) == 1\n",
    "    d['cities'] = d['cities'][0]\n",
    "\n",
    "    assert len(d['session_id']) == 1\n",
    "    d['session_id'] = d['session_id'][0]\n",
    "\n",
    "    assert len(d['unix_timestamp']) == 1\n",
    "    d['timestamp'] = datetime.datetime.utcfromtimestamp(d['unix_timestamp'][0])\n",
    "    del d['unix_timestamp']\n",
    "\n",
    "    # -------- retrieve users\n",
    "    user_dict = d['user']\n",
    "\n",
    "    assert len(user_dict) == 1\n",
    "    user_dict = user_dict[0]\n",
    "\n",
    "    assert len(user_dict) == 1\n",
    "    user_dict = user_dict[0]\n",
    "\n",
    "    d['user_id'] = user_dict['user_id']\n",
    "    d['user_country'] = user_dict['country']\n",
    "\n",
    "    del d['user']\n",
    "    return d"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "e5b0bdc8",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import datetime\n",
    "\n",
    "with open(\"city_search.json\",'rt') as inf:\n",
    "    sessions = json.load(inf)\n",
    "    \n",
    "for d in sessions:\n",
    "    clean_json(d)\n",
    "\n",
    "data = pd.DataFrame(sessions)\n",
    "data = data.set_index('session_id')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "ad47cf6a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Index: 20022 entries, X061RFWB06K9V to TVJS9Z6EZDPEA\n",
      "Data columns (total 4 columns):\n",
      " #   Column        Non-Null Count  Dtype         \n",
      "---  ------        --------------  -----         \n",
      " 0   cities        20022 non-null  object        \n",
      " 1   timestamp     20022 non-null  datetime64[ns]\n",
      " 2   user_id       20022 non-null  int64         \n",
      " 3   user_country  20022 non-null  object        \n",
      "dtypes: datetime64[ns](1), int64(1), object(2)\n",
      "memory usage: 782.1+ KB\n"
     ]
    }
   ],
   "source": [
    "data.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "efdfd593",
   "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>cities</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>user_country</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>session_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>X061RFWB06K9V</th>\n",
       "      <td>New York NY, Newark NJ</td>\n",
       "      <td>2015-09-17 15:28:28</td>\n",
       "      <td>2024</td>\n",
       "      <td>UK</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5AZ2X2A9BHH5U</th>\n",
       "      <td>New York NY, Jersey City NJ, Philadelphia PA</td>\n",
       "      <td>2015-09-04 08:06:31</td>\n",
       "      <td>2853</td>\n",
       "      <td>DE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SHTB4IYAX4PX6</th>\n",
       "      <td>San Antonio TX</td>\n",
       "      <td>2015-08-29 10:18:10</td>\n",
       "      <td>10958</td>\n",
       "      <td>UK</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>JBRB8MZGTX3M4</th>\n",
       "      <td>Edmonton AB</td>\n",
       "      <td>2015-03-25 07:21:03</td>\n",
       "      <td>7693</td>\n",
       "      <td>IT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>YJCMPURC2FL9C</th>\n",
       "      <td>Phoenix AZ, Houston TX</td>\n",
       "      <td>2015-05-02 09:31:07</td>\n",
       "      <td>7506</td>\n",
       "      <td>UK</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                     cities  \\\n",
       "session_id                                                    \n",
       "X061RFWB06K9V                        New York NY, Newark NJ   \n",
       "5AZ2X2A9BHH5U  New York NY, Jersey City NJ, Philadelphia PA   \n",
       "SHTB4IYAX4PX6                                San Antonio TX   \n",
       "JBRB8MZGTX3M4                                   Edmonton AB   \n",
       "YJCMPURC2FL9C                        Phoenix AZ, Houston TX   \n",
       "\n",
       "                        timestamp  user_id user_country  \n",
       "session_id                                               \n",
       "X061RFWB06K9V 2015-09-17 15:28:28     2024           UK  \n",
       "5AZ2X2A9BHH5U 2015-09-04 08:06:31     2853           DE  \n",
       "SHTB4IYAX4PX6 2015-08-29 10:18:10    10958           UK  \n",
       "JBRB8MZGTX3M4 2015-03-25 07:21:03     7693           IT  \n",
       "YJCMPURC2FL9C 2015-05-02 09:31:07     7506           UK  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0ab2c67d",
   "metadata": {},
   "source": [
    "### Part 2: There was a bug in the code and one country didn't get logged. It just shows up as an empty field (\"\"). Can you guess which country was that? How?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "6e3929ea",
   "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>country</th>\n",
       "      <th>user_country</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>US</td>\n",
       "      <td>3876</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>DE</td>\n",
       "      <td>3638</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>UK</td>\n",
       "      <td>3555</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td></td>\n",
       "      <td>2820</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>FR</td>\n",
       "      <td>2298</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>ES</td>\n",
       "      <td>1953</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>IT</td>\n",
       "      <td>1882</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  country  user_country\n",
       "0      US          3876\n",
       "1      DE          3638\n",
       "2      UK          3555\n",
       "3                  2820\n",
       "4      FR          2298\n",
       "5      ES          1953\n",
       "6      IT          1882"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check the value country\n",
    "data['user_country'].value_counts().reset_index().rename(columns={'index': 'country', 'country': 'count'})"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a6649aa6",
   "metadata": {},
   "source": [
    "There is a missing value in country but we cannot see it from the table above, therefore I will try to use other information to see if I can guess the country"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "79186ab2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# extract search time information\n",
    "data['time'] = pd.to_datetime(data['timestamp'], unit='s')\n",
    "data['hour'] = data['time'].apply(lambda x: x.hour)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "e21609e7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABQgAAANYCAYAAACIPgCjAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAACIc0lEQVR4nOz9fbRkd1knfH8v07yjA5iAkAQbmMjwMhAgE2EYGF5UQsyT8BYJN3AHxYk6QQKDSqIzMujKvRhBFBzUQch0Rl4DBImRl+QOgs7zEDDBIAkhQ4QATWISRMCRNWDgev6o3ebQOadT3af67K5dn89avar2rjrnfK9Vp0/99lW/397V3QEAAAAAVtP3jB0AAAAAABiPBiEAAAAArDANQgAAAABYYRqEAAAAALDCNAgBAAAAYIVtGzvAohx88MG9ffv2sWMAAAAAwAHp0ksv/XJ3H7L7/sk0CLdv355LLrlk7BgAAAAAcECqqs+vt98SYwAAAABYYRqEAAAAALDCJrPEGAAAAPbFe97x5bEjzO2EEw8eOwIwQWYQAgAAAMAK29IZhFV1UJJLknypu4+rqrsleXuS7UmuSfIT3f13w3PPSPL8JN9O8sLu/sBWZgUAAAAOLFf/zvVjR5jbP//5e4wdAea21TMIT0ty5Zrt05Nc1N1HJLlo2E5VPTDJSUkelOSYJL87NBcBAAAAgAXasgZhVR2W5MeTvGHN7hOSnD3cPzvJU9bsf1t3f7O7P5fk6iRHb1FUAAAAAFgZWzmD8LeT/FKS76zZd4/uvi5Jhtu7D/sPTfLFNc/bOez7LlV1SlVdUlWX3HjjjfslNAAAAABM2ZY0CKvquCQ3dPel837JOvv6Fju6X9/dR3X3UYcccsimMgIAAADAKtqqi5Q8OsnxVXVsktsn+b6qelOS66vqnt19XVXdM8kNw/N3Jjl8zdcfluTaLcoKAAAAACtjS2YQdvcZ3X1Yd2/P7OIjH+zu5yQ5L8nJw9NOTvKe4f55SU6qqttV1X2SHJHkY1uRFQAAAABWyVbNINzIK5KcU1XPT/KFJCcmSXdfUVXnJPlUkpuSnNrd3x4vJgAAACyPj5y9POfpf9TJThkGY9vyBmF3fyjJh4b7f5vkiRs878wkZ25ZMAAAAABYQWPPIAQAAABgYq5/zcVjR5jbPU575NgRRqdBCAAAADCSv/nNz4wdYW4/8JIjxo7AfqJBCAAAwFxe8+6/GTvC3E576g+MHQGYmBv+6/vGjjC3u7/gyXv1/C25ijEAAAAAcGDSIAQAAACAFaZBCAAAAAArTIMQAAAAAFaYBiEAAAAArDANQgAAAABYYRqEAAAAALDCNAgBAAAAYIVpEAIAAADACtMgBAAAAIAVpkEIAAAAACtMgxAAAAAAVti2sQMAAABM0TPP/ezYEeb29qfdd+wIAIzIDEIAAAAAWGEahAAAAACwwjQIAQAAAGCFaRACAAAAwArbp4uUVNWPJnlwko929/9vsZEAAAAAgK1yqzMIq+qtVfXTa7ZfmuT8JP9Xkgur6rn7MR8AAAAAsB/Ns8T40UnOS5Kq+p4kv5Dk/+ruf5XkGcM2AAAAALCE5mkQ3qW7bxjuPyzJ7ZP80bD9/iQ/uB9yAQAAAABbYJ4G4Zeravtw//FJPtLd3x6275Tk2+t+FQAAAABwwJvnIiVvSPInVfWBJP93kp9f89hjk1y5P4IBAAAAAPvfrc4g7O7/J8lvJLlNktO6+61rHj4kyW/e2veoqsOr6k+r6sqquqKqThv2362qLqyqzwy3d13zNWdU1dVVdVVVPWmvKwMAAAAAbtU8MwjT3WcnOXuD/fO4KclLuvvjVfW9SS6tqguTPC/JRd39iqo6PcnpSV5aVQ9MclKSByW5V5L/t6p+aM3SZgAAAABgAW61QVhVf5ikd9v9j0k+n+Qd3X2rS4y7+7ok1w33/76qrkxyaJITkjxueNrZST6U5KXD/rd19zeTfK6qrk5ydJKP3HpJAAAAAMC85plBePU6+26T5P5JLq6q/6u7/2TeHzhc8ORhST6a5B5D8zDdfV1V3X142qFJLl7zZTuHfbt/r1OSnJIk9773veeNAAAAAAAMbrVB2N0v3+ixqvqRJK9IMleDsKrunORdSV7U3V+vqg2ful6UdbK9Psnrk+Soo466xeMAAAAAwJ7d6kVKbsVFSe43zxOr6jaZNQff3N3nDruvr6p7Do/fM8kNw/6dSQ5f8+WHJbl2k1kBAAAAgN3MdZGSPbhXkq/e2pNqNlXwjUmu7O5Xr3novCQnZzYL8eQk71mz/y1V9erhZxyR5GP7EvDG33vTvnzZKA75ueeMHQEAALbc0961PKcaP/fpjxo7AgAs3DwXKbnvOrtvk2R7kv+Y5Jw5fs6jkzw3ySer6rJh3y9n1hg8p6qen+QLSU5Mku6+oqrOSfKpzK6AfKorGAMAAADA4s17kZLOd58X8NuZNfTenuTXbu0bdPf/zPrnFUySJ27wNWcmOXOOfAAAAADAPprnIiWbPU8hAAAAAHCA0vwDAAAAgBWmQQgAAAAAK0yDEAAAAABWmAYhAAAAAKwwDUIAAAAAWGEahAAAAACwwjQIAQAAAGCFaRACAAAAwArTIAQAAACAFbZt7ADsvet/7zfGjjC3e/zcL40dAQAAAIA9MIMQAAAAAFaYGYQAALBETnjn+8eOMJf3POOYsSMAAHMygxAAAAAAVpgGIQAAAACsMA1CAAAAAFhhGoQAAAAAsMI0CAEAAABghWkQAgAAAMAK0yAEAAAAgBWmQQgAAAAAK2zb2AEAAGB/OO6d7xg7wtzOf8aJY0cAAFaYBiEHjGte+5SxI8xt+wv/aOwIAAAAAAthiTEAAAAArDANQgAAAABYYRqEAAAAALDCDuhzEFbVMUlek+SgJG/o7leMHAkAYJKOe9eOsSPM5fynP2/sCAAAk3PANgir6qAkr0vyo0l2JvmLqjqvuz81bjKY30f/23FjR5jbD//M+WNHAAAAAEZwwDYIkxyd5Oru/mySVNXbkpyQRIMQRva+Nx47doS5Pfn57x07AjAxP37ua8eOMLc/edoLx44AAMASqO4eO8O6quoZSY7p7p8etp+b5Ie7+wVrnnNKklOGzfsnuWqL4h2c5Mtb9LO20hTrmmJNyTTrmmJNyTTrmmJNyTTrmmJNyTTrmmJNyTTrmmJNibqWyRRrSqZZ1xRrSqZZ1xRrSqZZ1xRrSra2rh/s7kN233kgzyCsdfZ9Vzezu1+f5PVbE+dmVXVJdx+11T93f5tiXVOsKZlmXVOsKZlmXVOsKZlmXVOsKZlmXVOsKZlmXVOsKVHXMpliTck065piTck065piTck065piTcmBUdeBfBXjnUkOX7N9WJJrR8oCAAAAAJN0IDcI/yLJEVV1n6q6bZKTkpw3ciYAAAAAmJQDdolxd99UVS9I8oEkByU5q7uvGDnWLlu+rHmLTLGuKdaUTLOuKdaUTLOuKdaUTLOuKdaUTLOuKdaUTLOuKdaUqGuZTLGmZJp1TbGmZJp1TbGmZJp1TbGm5ACo64C9SAkAAAAAsP8dyEuMAQAAAID9TIMQAAAAAFaYBuFeqqpjquqqqrq6qk4fO88iVNVZVXVDVV0+dpZFqarDq+pPq+rKqrqiqk4bO9NmVdXtq+pjVfWJoaaXj51pUarqoKr6y6o6f+wsi1JV11TVJ6vqsqq6ZOw8i1JVd6mqd1bVp4f/X48aO9NmVNX9h9do17+vV9WLxs61CFX14uFvxeVV9daquv3YmTarqk4b6rlimV+n9d53q+puVXVhVX1muL3rmBn3xQZ1nTi8Xt+pqqPGzLcvNqjplcPfwL+qqndX1V1GjLhPNqjr14eaLquqC6rqXmNm3Ft7Gs9W1S9UVVfVwWNk24wNXqv/XFVfWvPedeyYGffWRq9VVf38cJx1RVX9xlj59tUGr9Xb17xO11TVZSNG3Gsb1HRkVV28a4xbVUePmXFfbFDXQ6vqI8P4/Y+r6vvGzLi3Njr+XfbxxR7qWtrxxR5qGn18oUG4F6rqoCSvS/LkJA9M8qyqeuC4qRZiR5Jjxg6xYDcleUl3PyDJI5OcOoHX6ptJntDdD01yZJJjquqR40ZamNOSXDl2iP3g8d19ZHcv1ZvWrXhNkvd3979I8tAs+evW3VcNr9GRSR6R5BtJ3j1uqs2rqkOTvDDJUd394Mwu9nXSuKk2p6oenOTfJTk6s9+946rqiHFT7bMdueX77ulJLuruI5JcNGwvmx25ZV2XJ3lakj/b8jSLsSO3rOnCJA/u7ock+V9JztjqUAuwI7es65Xd/ZDh7+H5SX51q0Nt0o6sM56tqsOT/GiSL2x1oAXZkfXH6b+16/2ru9+7xZk2a0d2q6mqHp/khCQP6e4HJXnVCLk2a0d2q6u7n7lmnPGuJOeOkGszduSWv3+/keTlQ02/Omwvmx25ZV1vSHJ6d//LzMaCv7jVoTZpo+PfZR9fbFTXMo8vNqpp9PGFBuHeOTrJ1d392e7+VpK3ZfZGttS6+8+SfGXsHIvU3dd198eH+3+fWRPj0HFTbU7P/O9h8zbDv6W/ylBVHZbkxzN7U+YANnyS+tgkb0yS7v5Wd3911FCL9cQkf93dnx87yIJsS3KHqtqW5I5Jrh05z2Y9IMnF3f2N7r4pyYeTPHXkTPtkg/fdE5KcPdw/O8lTtjLTIqxXV3df2d1XjRRp0zao6YLhdzBJLk5y2JYH26QN6vr6ms07ZcnGGHsYz/5Wkl/KktWzy0TH6evV9HNJXtHd3xyec8OWB9ukPb1WVVVJfiLJW7c01CZtUFMn2TW77p9lCccXG9R1/9zcbLowydO3NNQm7eH4d6nHFxvVtczjiz3UNPr4QoNw7xya5ItrtndmyZtOq6Cqtid5WJKPjhxl02q2FPeyJDckubC7l76mJL+d2cD9OyPnWLROckFVXVpVp4wdZkHum+TGJP+9ZkvC31BVdxo71AKdlCUbuG+ku7+U2eyLLyS5LsnXuvuCcVNt2uVJHltV319Vd0xybJLDR860SPfo7uuS2cAxyd1HzsN8firJ+8YOsShVdWZVfTHJs7N8MwhvoaqOT/Kl7v7E2Fn2gxcMy9DOWrYlgxv4oSSPqaqPVtWHq+pfjR1owR6T5Pru/szYQRbgRUleOfyteFWWcxb1ei5Pcvxw/8Qs8Rhjt+PfyYwvpnRcv8seahplfKFBuHdqnX1L+WnkqqiqO2c2nf9Fu30yvpS6+9vDdP7Dkhw9LLlbWlV1XJIbuvvSsbPsB4/u7odndkqCU6vqsWMHWoBtSR6e5Pe6+2FJ/iHLt0xhXVV128wGhe8YO8siDAeLJyS5T5J7JblTVT1n3FSb091XJvkvmX2q//4kn8hsiQaMoqp+JbPfwTePnWVRuvtXuvvwzGp6wdh5NmP4IOFXMoFG5zp+L8n9MjvlzHVJfnPUNIuxLcldM1tu94tJzhlm3U3FszKRDyEzm+354uFvxYszrCyZgJ/KbMx+aZLvTfKtkfPsk6kd/+4yxbo2qmnM8YUG4d7Zme/+JOGwLOGU6lVRVbfJ7D/cm7t72c73sUfDss4PZfnPHfnoJMdX1TWZLdl/QlW9adxIi9Hd1w63N2R2HpOlO4HzOnYm2blm5uo7M2sYTsGTk3y8u68fO8iC/EiSz3X3jd39j5md8+hfj5xp07r7jd398O5+bGZLg6YwE2OX66vqnkky3C7d8rpVUlUnJzkuybO7e4ofFr8lS7a8bh33y+xDkk8M44zDkny8qn5g1FQL0N3XDx8afyfJH2Q6Y4xzh1PqfCyzlSVLd1GZ9Qyn+nhakrePnWVBTs7N51J8R6bx+5fu/nR3/1h3PyKzZu5fj51pb21w/Lv044spHtdvVNPY4wsNwr3zF0mOqKr7DLNNTkpy3siZWMfwieMbk1zZ3a8eO88iVNUhu65kVFV3yKwB8OlRQ21Sd5/R3Yd19/bM/j99sLuXepZTklTVnarqe3fdT/JjmS1bWGrd/TdJvlhV9x92PTHJp0aMtEhT+mQ/mS0tfmRV3XH4e/jELPkFZZKkqu4+3N47s4OtKb1m52V20JXh9j0jZmEPquqYJC9Ncnx3f2PsPIuy20V/js/yjzE+2d137+7twzhjZ5KHD+9lS23Xwf7gqZnAGCPJHyV5QpJU1Q8luW2SL48ZaIF+JMmnu3vn2EEW5Nok/3a4/4RM5MO6NWOM70nyH5P8/riJ9s4ejn+Xenwx0eP6dWs6EMYXNc0PPfefqjo2s3OmHZTkrO4+c9xEm1dVb03yuMw+pbs+ycu6e6mnilfVv0ny50k+mZvPbffLvXxXefsnVfWQzE4se1Bmzf1zuvvXxk21OFX1uCS/0N3HjRxl06rqvrn5SrjbkrxlCn8rkqSqjszsgjK3TfLZJD/Z3X83aqhNGpahfTHJfbv7a2PnWZSqenmSZ2a2ROEvk/z0rpO/L6uq+vMk35/kH5P8h+6+aORI+2S9993MDo7PSXLvzBq8J3b3Ul2YYIO6vpLkd5IckuSrSS7r7ieNFHGvbVDTGUlul+Rvh6dd3N0/O0rAfbRBXcdmdpL+7yT5fJKfHc5nuhRubTw7zCI8qruXqum0wWv1uMyWF3eSa5L8zK5zjC2DDWr6wyRnZVbXtzIbE35wpIj7ZKPfwarakdnfiaVqOCUbvlZXJXlNZmPc/5Pk3y/b6YI2qOvOSU4dnnJukjOWaYb4Rse/mZ3bbmnHF3uo63ZZ0vHFHmp6bUYeX2gQAgAAAMAKs8QYAAAAAFaYBiEAAAAArDANQgAAAABYYRqEAAAAALDCNAgBAAAAYIVpEAIAMLeq2l5Vl4+dAwCAxdEgBABgVFW1bewMAACrTIMQAIC9dVBV/UFVXVFVF1TVHarqyKq6uKr+qqreXVV3TZKq+lBVHTXcP7iqrhnuP6+q3lFVf5zkgvFKAQBAgxAAgL11RJLXdfeDknw1ydOT/I8kL+3uhyT5ZJKXzfF9HpXk5O5+wv4KCgDArdMgBABgb32uuy8b7l+a5H5J7tLdHx72nZ3ksXN8nwu7+yv7IR8AAHtBgxAAgL31zTX3v53kLnt47k25ecx5+90e+4cFZgIAYB9pEAIAsFlfS/J3VfWYYfu5SXbNJrwmySOG+8/Y4lwAAMzBFeMAAFiEk5P8flXdMclnk/zksP9VSc6pqucm+eBY4QAA2Fh199gZAAAAAICRWGIMAAAAACtMgxAAAAAAVpgGIQAAAACsMA1CAAAAAFhhGoQAAAAAsMI0CAEAAABghWkQAgAAAMAK0yAEAAAAgBWmQQgAAAAAK0yDEAAAAABWmAYhAAAAAKwwDUIAAAAAWGFb2iCsqoOq6i+r6vxh+25VdWFVfWa4veua555RVVdX1VVV9aStzAkAAAAAq6K6e+t+WNV/SHJUku/r7uOq6jeSfKW7X1FVpye5a3e/tKoemOStSY5Ocq8k/2+SH+rub2/0vQ8++ODevn37/i8CAAAAAJbQpZde+uXuPmT3/du2KkBVHZbkx5OcmeQ/DLtPSPK44f7ZST6U5KXD/rd19zeTfK6qrs6sWfiRjb7/9u3bc8kll+yX7AAAAACw7Krq8+vt38olxr+d5JeSfGfNvnt093VJMtzefdh/aJIvrnnezmHfd6mqU6rqkqq65MYbb9wvoQEAAABgyrZkBmFVHZfkhu6+tKoeN8+XrLPvFmuhu/v1SV6fJEcdddTWrZUGgAV78nnHjx1hbu87/ry5nnfsu1+2n5Msznuf+vKxIwAAwGi2aonxo5McX1XHJrl9ku+rqjclub6q7tnd11XVPZPcMDx/Z5LD13z9YUmu3aKsAABwwDrhne8fO8Jc3vOMY8aOAADMaUuWGHf3Gd19WHdvT3JSkg9293OSnJfk5OFpJyd5z3D/vCQnVdXtquo+SY5I8rGtyAoAAAAAq2TLLlKygVckOaeqnp/kC0lOTJLuvqKqzknyqSQ3JTl1T1cwBgAAAAD2zZY3CLv7Q5ldrTjd/bdJnrjB887M7IrHAAAAAMB+spVXMQYAAAAADjAahAAAAACwwsY+ByEAAAeA4961Y+wIczn/6c8bOwIAwOSYQQgAAAAAK0yDEAAAAABWmCXGAABM0nHvfMfYEeZ2/jNOHDsCALDCzCAEAAAAgBVmBiEAwF748XNfO3aEuf3J0144dgSYy9Pe9ZGxI8zt3Kc/auwIALBwZhACAAAAwArTIAQAAACAFaZBCAAAAAArTIMQAAAAAFaYBiEAAAAArDANQgAAAABYYRqEAAAAALDCNAgBAAAAYIVpEAIAAADACtMgBAAAAIAVtm3sAACwN/7TOceMHWFuv/4T7x87AgAjeua5nx07wtze/rT7jh0BgBGZQQgAAAAAK0yDEAAAAABWmAYhAAAAAKwwDUIAAAAAWGEahAAAAACwwjQIAQAAAGCFaRACAAAAwArTIAQAAACAFXarDcKq+svdtv/7bts3LDoUAAAAALA15plB+M932z5ht+07LCgLAAAAALDF5mkQ9m7bdSuP30JVHV5Vf1pVV1bVFVV12rD/blV1YVV9Zri965qvOaOqrq6qq6rqSXPkBAAAAAD20r6cg/BWG4LruCnJS7r7AUkemeTUqnpgktOTXNTdRyS5aNjO8NhJSR6U5Jgkv1tVB+3DzwUAAAAA9mDbHM+5XVX92prtO+y2fdtb+wbdfV2S64b7f19VVyY5NLPlyo8bnnZ2kg8leemw/23d/c0kn6uqq5McneQjc+QFAAAAAOY0T4PwLUkOX7P9tt2237o3P7Cqtid5WJKPJrnH0DxMd19XVXcfnnZokovXfNnOYd/u3+uUJKckyb3vfe+9iQEAAAAAZI4GYXf/ZPJPy34fk+RuSb6S5H929xV788Oq6s5J3pXkRd399ardT2d481PXi7JOttcneX2SHHXUUfuy9BkAAAAAVtqtNghr1sV7Q5KTk3wpybWZzea7V1X9YZKf6u55LlRym8yag2/u7nOH3ddX1T2H2YP3THLDsH9nvnuW4mHDzwUAAAAAFmiei5SckuTxSR7V3T/Y3Y/q7nsneVRmMwp/5ta+wdBkfGOSK7v71WseOi+zxmOG2/es2X9SVd2uqu6T5IgkH5unIAAAAABgfvM0CJ+b5IXd/Rdrdw7bLxoevzWPHp73hKq6bPh3bJJXJPnRqvpMkh8dtjMsXT4nyaeSvD/Jqd397flKAgAAAADmNc9FSh6Y5MMbPPbhJH94a9+gu/9n1j+vYJI8cYOvOTPJmXPkAwAAAAD20TwNwoO6++/Xe6C7/76q5pmFCAAAwJJ7zbv/ZuwIczvtqT8wdgSApTFPg/A2VfX4bDwDcJ7vAQAAAAAcgOZp7t2Q5KxbeRwAAAAAWEK32iDs7u1bkAOA/eC/vulJY0eY2wue84GxIwAAAKwk5w8EAAAAgBWmQQgAAAAAK0yDEAAAAABWmAYhAAAAAKwwDUIAAAAAWGEahAAAAACwwjQIAQAAAGCFaRACAAAAwArTIAQAAACAFaZBCAAAAAArTIMQAAAAAFaYBiEAAAAArDANQgAAAABYYRqEAAAAALDCto0dAOBA8OYdTxo7wtye/bwPjB0BAGBS3vOOL48dYW4nnHjw2BGACTKDEAAAAABWmAYhAAAAAKwwDUIAAAAAWGEahAAAAACwwjQIAQAAAGCFaRACAAAAwArTIAQAAACAFbZt7AAAAADAYn3k7BvHjjC3R518yNgRYOVpEAJ77X1vPHbsCHN78vPfO3YEAAAAOKBZYgwAAAAAK+yAbhBW1TFVdVVVXV1Vp4+dBwAAAACm5oBdYlxVByV5XZIfTbIzyV9U1Xnd/alxk8H8Pvrfjhs7wtx++GfOHzsCAADAHl39O9ePHWFu//zn7zF2BJjbAdsgTHJ0kqu7+7NJUlVvS3JCEg1CAAAAYBL+5jc/M3aEuf3AS44YOwL7SXX32BnWVVXPSHJMd//0sP3cJD/c3S9Y85xTkpwybN4/yVVbFO/gJF/eop+1laZY1xRrSqZZ1xRrSqZZ1xRrSqZZ1xRrSqZZ1xRrSqZZ1xRrStS1TKZYUzLNuqZYUzLNuqZYUzLNuqZYU7K1df1gd9/i0uEH8gzCWmffd3Uzu/v1SV6/NXFuVlWXdPdRW/1z97cp1jXFmpJp1jXFmpJp1jXFmpJp1jXFmpJp1jXFmpJp1jXFmhJ1LZMp1pRMs64p1pRMs64p1pRMs64p1pQcGHUdyBcp2Znk8DXbhyW5dqQsAAAAADBJB3KD8C+SHFFV96mq2yY5Kcl5I2cCAAAAgEk5YJcYd/dNVfWCJB9IclCSs7r7ipFj7bLly5q3yBTrmmJNyTTrmmJNyTTrmmJNyTTrmmJNyTTrmmJNyTTrmmJNibqWyRRrSqZZ1xRrSqZZ1xRrSqZZ1xRrSg6Aug7Yi5QAAAAAAPvfgbzEGAAAAADYzzQIAQAAAGCFaRDupao6pqquqqqrq+r0sfMsQlWdVVU3VNXlY2dZlKo6vKr+tKqurKorquq0sTNtVlXdvqo+VlWfGGp6+diZFqWqDqqqv6yq88fOsihVdU1VfbKqLquqS8bOsyhVdZeqemdVfXr4//WosTNtRlXdf3iNdv37elW9aOxci1BVLx7+VlxeVW+tqtuPnWmzquq0oZ4rlvl1Wu99t6ruVlUXVtVnhtu7jplxX2xQ14nD6/WdqjpqzHz7YoOaXjn8Dfyrqnp3Vd1lxIj7ZIO6fn2o6bKquqCq7jVmxr21p/FsVf1CVXVVHTxGts3Y4LX6z1X1pTXvXceOmXFvbfRaVdXPD8dZV1TVb4yVb19t8Fq9fc3rdE1VXTZixL22QU1HVtXFu8a4VXX0mBn3xQZ1PbSqPjKM3/+4qr5vzIx7a6Pj32UfX+yhrqUdX+yhptHHFxqEe6GqDkryuiRPTvLAJM+qqgeOm2ohdiQ5ZuwQC3ZTkpd09wOSPDLJqRN4rb6Z5And/dAkRyY5pqoeOW6khTktyZVjh9gPHt/dR3b3Ur1p3YrXJHl/d/+LJA/Nkr9u3X3V8BodmeQRSb6R5N3jptq8qjo0yQuTHNXdD87sYl8njZtqc6rqwUn+XZKjM/vdO66qjhg31T7bkVu+756e5KLuPiLJRcP2stmRW9Z1eZKnJfmzLU+zGDtyy5ouTPLg7n5Ikv+V5IytDrUAO3LLul7Z3Q8Z/h6en+RXtzrUJu3IOuPZqjo8yY8m+cJWB1qQHVl/nP5bu96/uvu9W5xps3Zkt5qq6vFJTkjykO5+UJJXjZBrs3Zkt7q6+5lrxhnvSnLuCLk2Y0du+fv3G0lePtT0q8P2stmRW9b1hiSnd/e/zGws+ItbHWqTNjr+XfbxxUZ1LfP4YqOaRh9faBDunaOTXN3dn+3ubyV5W2ZvZEutu/8syVfGzrFI3X1dd398uP/3mTUxDh031eb0zP8eNm8z/Fv6qwxV1WFJfjyzN2UOYMMnqY9N8sYk6e5vdfdXRw21WE9M8tfd/fmxgyzItiR3qKptSe6Y5NqR82zWA5Jc3N3f6O6bknw4yVNHzrRPNnjfPSHJ2cP9s5M8ZSszLcJ6dXX3ld191UiRNm2Dmi4YfgeT5OIkh215sE3aoK6vr9m8U5ZsjLGH8exvJfmlLFk9u0x0nL5eTT+X5BXd/c3hOTdsebBN2tNrVVWV5CeSvHVLQ23SBjV1kl2z6/5ZlnB8sUFd98/NzaYLkzx9S0Nt0h6Of5d6fLFRXcs8vthDTaOPLzQI986hSb64ZntnlrzptAqqanuShyX56MhRNq1mS3EvS3JDkgu7e+lrSvLbmQ3cvzNyjkXrJBdU1aVVdcrYYRbkvkluTPLfa7Yk/A1VdaexQy3QSVmygftGuvtLmc2++EKS65J8rbsvGDfVpl2e5LFV9f1VdcckxyY5fORMi3SP7r4umQ0ck9x95DzM56eSvG/sEItSVWdW1ReTPDvLN4PwFqrq+CRf6u5PjJ1lP3jBsAztrGVbMriBH0rymKr6aFV9uKr+1diBFuwxSa7v7s+MHWQBXpTklcPfildlOWdRr+fyJMcP90/MEo8xdjv+ncz4YkrH9bvsoaZRxhcahHun1tm3lJ9GroqqunNm0/lftNsn40upu789TOc/LMnRw5K7pVVVxyW5obsvHTvLfvDo7n54ZqckOLWqHjt2oAXYluThSX6vux+W5B+yfMsU1lVVt81sUPiOsbMswnCweEKS+yS5V5I7VdVzxk21Od19ZZL/ktmn+u9P8onMlmjAKKrqVzL7HXzz2FkWpbt/pbsPz6ymF4ydZzOGDxJ+JRNodK7j95LcL7NTzlyX5DdHTbMY25LcNbPldr+Y5Jxh1t1UPCsT+RAys9meLx7+Vrw4w8qSCfipzMbslyb53iTfGjnPPpna8e8uU6xro5rGHF9oEO6dnfnuTxIOyxJOqV4VVXWbzP7Dvbm7l+18H3s0LOv8UJb/3JGPTnJ8VV2T2ZL9J1TVm8aNtBjdfe1we0Nm5zFZuhM4r2Nnkp1rZq6+M7OG4RQ8OcnHu/v6sYMsyI8k+Vx339jd/5jZOY/+9ciZNq2739jdD+/ux2a2NGgKMzF2ub6q7pkkw+3SLa9bJVV1cpLjkjy7u6f4YfFbsmTL69Zxv8w+JPnEMM44LMnHq+oHRk21AN19/fCh8XeS/EGmM8Y4dzilzscyW1mydBeVWc9wqo+nJXn72FkW5OTcfC7Fd2Qav3/p7k9394919yMya+b+9diZ9tYGx79LP76Y4nH9RjWNPb7QINw7f5HkiKq6zzDb5KQk542ciXUMnzi+McmV3f3qsfMsQlUdsutKRlV1h8waAJ8eNdQmdfcZ3X1Yd2/P7P/TB7t7qWc5JUlV3amqvnfX/SQ/ltmyhaXW3X+T5ItVdf9h1xOTfGrESIs0pU/2k9nS4kdW1R2Hv4dPzJJfUCZJquruw+29MzvYmtJrdl5mB10Zbt8zYhb2oKqOSfLSJMd39zfGzrMou1305/gs/xjjk9199+7ePowzdiZ5+PBettR2HewPnpoJjDGS/FGSJyRJVf1Qktsm+fKYgRboR5J8urt3jh1kQa5N8m+H+0/IRD6sWzPG+J4k/zHJ74+baO/s4fh3qccXEz2uX7emA2F8UdP80HP/qapjMztn2kFJzuruM8dNtHlV9dYkj8vsU7rrk7ysu5d6qnhV/Zskf57kk7n53Ha/3Mt3lbd/UlUPyezEsgdl1tw/p7t/bdxUi1NVj0vyC9193MhRNq2q7pubr4S7LclbpvC3Ikmq6sjMLihz2ySfTfKT3f13o4bapGEZ2heT3Le7vzZ2nkWpqpcneWZmSxT+MslP7zr5+7Kqqj9P8v1J/jHJf+jui0aOtE/We9/N7OD4nCT3zqzBe2J3L9WFCTao6ytJfifJIUm+muSy7n7SSBH32gY1nZHkdkn+dnjaxd39s6ME3Ecb1HVsZifp/06Szyf52eF8pkvh1sazwyzCo7p7qZpOG7xWj8tseXEnuSbJz+w6x9gy2KCmP0xyVmZ1fSuzMeEHR4q4Tzb6HayqHZn9nViqhlOy4Wt1VZLXZDbG/T9J/v2ynS5og7runOTU4SnnJjljmWaIb3T8m9m57ZZ2fLGHum6XJR1f7KGm12bk8YUGIQAAAACsMEuMAQAAAGCFaRACAAAAwArTIAQAAACAFaZBCAAAAAArTIMQAAAAAFaYBiEAAHOrqu1VdfnYOQAAWBwNQgAARlVV28bOAACwyjQIAQDYWwdV1R9U1RVVdUFV3aGqjqyqi6vqr6rq3VV11ySpqg9V1VHD/YOr6prh/vOq6h1V9cdJLhivFAAANAgBANhbRyR5XXc/KMlXkzw9yf9I8tLufkiSTyZ52Rzf51FJTu7uJ+yvoAAA3DoNQgAA9tbnuvuy4f6lSe6X5C7d/eFh39lJHjvH97mwu7+yH/IBALAXNAgBANhb31xz/9tJ7rKH596Um8ect9/tsX9YYCYAAPaRBiEAAJv1tSR/V1WPGbafm2TXbMJrkjxiuP+MLc4FAMAcXDEOAIBFODnJ71fVHZN8NslPDvtfleScqnpukg+OFQ4AgI1Vd4+dAQAAAAAYiSXGAAAAALDCNAgBAAAAYIVpEAIAAADACtMgBAAAAIAVpkEIAAAAACtMgxAAAAAAVpgGIQAAAACsMA1CAAAAAFhhGoQAAAAAsMI0CAEAAABghWkQAgAAAMAK0yAEAAAAgBWmQQgAAAAAK2zb2AEW5eCDD+7t27ePHQMAAAAADkiXXnrpl7v7kN33T6ZBuH379lxyySVjxwAAAACAA1JVfX69/ZYYAwAAAMAKm8wMQgCArfDj57527Ahz+5OnvXDsCOwHJ7zz/WNHmMt7nnHM2BEAgDlpEAIAAKN62rs+MnaEuZ379EeNHQEAFk6DEAAAYD945rmfHTvC3N7+tPuOHQGAETkHIQAAAACsMA1CAAAAAFhhGoQAAAAAsMKcgxAAgBz3rh1jR5jL+U9/3tgRAAAmxwxCAAAAAFhhGoQAAAAAsMI0CAEAAABghTkHIQCwXxz77peNHWFu733qy8eOAAAAozGDEAAAAABWmBmEAABM0nHvfMfYEeZ2/jNOHDsCALDCzCAEAAAAgBWmQQgAAAAAK0yDEAAAAABWmAYhAAAAAKwwDUIAAAAAWGEahAAAAACwwjQIAQAAAGCFaRACAAAAwArbNnYAACB58nnHjx1hbu87/ryxIwAAAAtkBiEAAAAArDANQgAAAABYYRqEAAAAALDCNAgBAAAAYIVpEAIAAADACnMVYwAAAObymnf/zdgR5nbaU39g7AgAS8MMQgAAAABYYRqEAAAAALDCbrVBWFV3vpXHj1pcHAAAAABgK80zg/B9VXX79R6oqn+d5ILFRgIAAAAAtso8DcIbk5xfVbdbu7OqHpfkvUl+6da+QVUdXlV/WlVXVtUVVXXasP9uVXVhVX1muL3rmq85o6qurqqrqupJe1MUAAAAADCfeRqEz0zyf5K8u6pukyRV9WNJ3pPkhd39hjm+x01JXtLdD0jyyCSnVtUDk5ye5KLuPiLJRcN2hsdOSvKgJMck+d2qOmivKgMAAAAAbtWtNgi7+x+TPC3JtiTvqKqnJnlXkp/p7v8xzw/p7uu6++PD/b9PcmWSQ5OckOTs4WlnJ3nKcP+EJG/r7m929+eSXJ3k6HmLAgAAAADmM9dVjLv7W5k17e6S5K1Jntvdb9uXH1hV25M8LMlHk9yju68bfsZ1Se4+PO3QJF9c82U7h327f69TquqSqrrkxhtv3Jc4AAAAALDS5rmK8Rer6gtJrkrywCSd5LVV9YVd/+b9YcMVkd+V5EXd/fU9PXWdfX2LHd2v7+6juvuoQw45ZN4YAAAAAMBg2xzPec4iftBw/sJ3JXlzd5877L6+qu7Z3ddV1T2T3DDs35nk8DVffliSaxeRAwAAAAC42a02CLv7w5v9IVVVSd6Y5MrufvWah85LcnKSVwy371mz/y1V9eok90pyRJKPbTYHAAAAAPDdbrVBWFXrXYjkH5N8Psk7u/tTc/ycRyd5bpJPVtVlw75fzqwxeE5VPT/JF5KcmCTdfUVVnZPkU5ldAfnU7v72HD8HAAAAANgL8ywx/ut19t0myf2TfKSq/q/u/pM9fYPu/p9Z/7yCSfLEDb7mzCRnzpEPAAAA9tl73vHlsSPM7YQTDx47AjBB8ywxfvlGj1XVj2Q2C3CPDUIAAAAA4MA0zwzCPbkoyf0WEQQA5vGfzjlm7Ahz+/WfeP/YEQAAAG7V92zy6++V5KsLyAEAAAAAjGCei5Tcd53dt0myPcl/THLOgjMBAAAAAFtkniXGVyfpfPdFRr6d2VWH357k1/ZDLgAAAABgC8xzkZLNLkMGAAAAAA5Qmn8AAAAAsMI0CAEAAABghWkQAgAAAMAK0yAEAAAAgBWmQQgAAAAAK0yDEAAAAABWmAYhAAAAAKwwDUIAAAAAWGEahAAAAACwwjQIAQAAAGCFaRACAAAAwArTIAQAAACAFaZBCAAAAAArTIMQAAAAAFbYtrEDALD//Nc3PWnsCHN7wXM+MHYEAIDJ+MjZN44dYW6POvmQsSPAyjODEAAAAABWmAYhAAAAAKwwDUIAAAAAWGEahAAAAACwwjQIAQAAAGCFaRACAAAAwArTIAQAAACAFbZt7AAAB4I373jS2BHm9uznfWDsCAAAAEyIGYQAAAAAsMLMIAT22vveeOzYEeb25Oe/d+wIAADAglz9O9ePHWFu//zn7zF2BJjbAT2DsKqOqaqrqurqqjp97DwAAAAAMDUHbIOwqg5K8rokT07ywCTPqqoHjpsKAAAAAKblQF5ifHSSq7v7s0lSVW9LckKST42aCvbCR//bcWNHmNsP/8z5Y0cAAABYOX/zm58ZO8LcfuAlR4wdgf2kunvsDOuqqmckOaa7f3rYfm6SH+7uF6x5zilJThk275/kqi2Kd3CSL2/Rz9pKU6xrijUl06xrijUl06xrijUl06xrijUl06xrijUl06xrijUl6lomU6wpmWZdU6wpmWZdU6wpmWZdU6wp2dq6frC7D9l954E8g7DW2fdd3czufn2S129NnJtV1SXdfdRW/9z9bYp1TbGmZJp1TbGmZJp1TbGmZJp1TbGmZJp1TbGmZJp1TbGmRF3LZIo1JdOsa4o1JdOsa4o1JdOsa4o1JQdGXQfsOQiT7Exy+Jrtw5JcO1IWAAAAAJikA7lB+BdJjqiq+1TVbZOclOS8kTMBAAAAwKQcsEuMu/umqnpBkg8kOSjJWd19xcixdtnyZc1bZIp1TbGmZJp1TbGmZJp1TbGmZJp1TbGmZJp1TbGmZJp1TbGmRF3LZIo1JdOsa4o1JdOsa4o1JdOsa4o1JQdAXQfsRUoAAAAAgP3vQF5iDAAAAADsZxqEAAAAALDCNAj3UlUdU1VXVdXVVXX62HkWoarOqqobqurysbMsSlUdXlV/WlVXVtUVVXXa2Jk2q6puX1Ufq6pPDDW9fOxMi1JVB1XVX1bV+WNnWZSquqaqPllVl1XVJWPnWZSquktVvbOqPj38/3rU2Jk2o6ruP7xGu/59vapeNHauRaiqFw9/Ky6vqrdW1e3HzrRZVXXaUM8Vy/w6rfe+W1V3q6oLq+ozw+1dx8y4Lzao68Th9fpOVR01Zr59sUFNrxz+Bv5VVb27qu4yYsR9skFdvz7UdFlVXVBV9xoz497a03i2qn6hqrqqDh4j22Zs8Fr956r60pr3rmPHzLi3Nnqtqurnh+OsK6rqN8bKt682eK3evuZ1uqaqLhsx4l7boKYjq+riXWPcqjp6zIz7YoO6HlpVHxnG739cVd83Zsa9tdHx77KPL/ZQ19KOL/ZQ0+jjCw3CvVBVByV5XZInJ3lgkmdV1QPHTbUQO5IcM3aIBbspyUu6+wFJHpnk1Am8Vt9M8oTufmiSI5McU1WPHDfSwpyW5MqxQ+wHj+/uI7t7qd60bsVrkry/u/9FkodmyV+37r5qeI2OTPKIJN9I8u5xU21eVR2a5IVJjuruB2d2sa+Txk21OVX14CT/LsnRmf3uHVdVR4ybap/tyC3fd09PclF3H5HkomF72ezILeu6PMnTkvzZlqdZjB25ZU0XJnlwdz8kyf9KcsZWh1qAHbllXa/s7ocMfw/PT/KrWx1qk3ZknfFsVR2e5EeTfGGrAy3Ijqw/Tv+tXe9f3f3eLc60WTuyW01V9fgkJyR5SHc/KMmrRsi1WTuyW13d/cw144x3JTl3hFybsSO3/P37jSQvH2r61WF72ezILet6Q5LTu/tfZjYW/MWtDrVJGx3/Lvv4YqO6lnl8sVFNo48vNAj3ztFJru7uz3b3t5K8LbM3sqXW3X+W5Ctj51ik7r6uuz8+3P/7zJoYh46banN65n8Pm7cZ/i39VYaq6rAkP57ZmzIHsOGT1McmeWOSdPe3uvuro4ZarCcm+evu/vzYQRZkW5I7VNW2JHdMcu3IeTbrAUku7u5vdPdNST6c5KkjZ9onG7zvnpDk7OH+2UmespWZFmG9urr7yu6+aqRIm7ZBTRcMv4NJcnGSw7Y82CZtUNfX12zeKUs2xtjDePa3kvxSlqyeXSY6Tl+vpp9L8oru/ubwnBu2PNgm7em1qqpK8hNJ3rqloTZpg5o6ya7Zdf8sSzi+2KCu++fmZtOFSZ6+paE2aQ/Hv0s9vtiormUeX+yhptHHFxqEe+fQJF9cs70zS950WgVVtT3Jw5J8dOQom1azpbiXJbkhyYXdvfQ1JfntzAbu3xk5x6J1kguq6tKqOmXsMAty3yQ3JvnvNVsS/oaqutPYoRbopCzZwH0j3f2lzGZffCHJdUm+1t0XjJtq0y5P8tiq+v6qumOSY5McPnKmRbpHd1+XzAaOSe4+ch7m81NJ3jd2iEWpqjOr6otJnp3lm0F4C1V1fJIvdfcnxs6yH7xgWIZ21rItGdzADyV5TFV9tKo+XFX/auxAC/aYJNd392fGDrIAL0ryyuFvxauynLOo13N5kuOH+ydmiccYux3/TmZ8MaXj+l32UNMo4wsNwr1T6+xbyk8jV0VV3Tmz6fwv2u2T8aXU3d8epvMfluToYcnd0qqq45Lc0N2Xjp1lP3h0dz88s1MSnFpVjx070AJsS/LwJL/X3Q9L8g9ZvmUK66qq22Y2KHzH2FkWYThYPCHJfZLcK8mdquo546banO6+Msl/yexT/fcn+URmSzRgFFX1K5n9Dr557CyL0t2/0t2HZ1bTC8bOsxnDBwm/kgk0Otfxe0nul9kpZ65L8pujplmMbUnumtlyu19Mcs4w624qnpWJfAiZ2WzPFw9/K16cYWXJBPxUZmP2S5N8b5JvjZxnn0zt+HeXKda1UU1jji80CPfOznz3JwmHZQmnVK+KqrpNZv/h3tzdy3a+jz0alnV+KMt/7shHJzm+qq7JbMn+E6rqTeNGWozuvna4vSGz85gs3Qmc17Ezyc41M1ffmVnDcAqenOTj3X392EEW5EeSfK67b+zuf8zsnEf/euRMm9bdb+zuh3f3YzNbGjSFmRi7XF9V90yS4Xbpltetkqo6OclxSZ7d3VP8sPgtWbLldeu4X2YfknxiGGccluTjVfUDo6ZagO6+fvjQ+DtJ/iDTGWOcO5xS52OZrSxZuovKrGc41cfTkrx97CwLcnJuPpfiOzKN379096e7+8e6+xGZNXP/euxMe2uD49+lH19M8bh+o5rGHl9oEO6dv0hyRFXdZ5htclKS80bOxDqGTxzfmOTK7n712HkWoaoO2XUlo6q6Q2YNgE+PGmqTuvuM7j6su7dn9v/pg9291LOckqSq7lRV37vrfpIfy2zZwlLr7r9J8sWquv+w64lJPjVipEWa0if7yWxp8SOr6o7D38MnZskvKJMkVXX34fbemR1sTek1Oy+zg64Mt+8ZMQt7UFXHJHlpkuO7+xtj51mU3S76c3yWf4zxye6+e3dvH8YZO5M8fHgvW2q7DvYHT80ExhhJ/ijJE5Kkqn4oyW2TfHnMQAv0I0k+3d07xw6yINcm+bfD/SdkIh/WrRljfE+S/5jk98dNtHf2cPy71OOLiR7Xr1vTgTC+qGl+6Ln/VNWxmZ0z7aAkZ3X3meMm2ryqemuSx2X2Kd31SV7W3Us9Vbyq/k2SP0/yydx8brtf7uW7yts/qaqHZHZi2YMya+6f092/Nm6qxamqxyX5he4+buQom1ZV983NV8LdluQtU/hbkSRVdWRmF5S5bZLPJvnJ7v67UUNt0rAM7YtJ7tvdXxs7z6JU1cuTPDOzJQp/meSnd538fVlV1Z8n+f4k/5jkP3T3RSNH2ifrve9mdnB8TpJ7Z9bgPbG7l+rCBBvU9ZUkv5PkkCRfTXJZdz9ppIh7bYOazkhyuyR/Ozzt4u7+2VEC7qMN6jo2s5P0fyfJ55P87HA+06Vwa+PZYRbhUd29VE2nDV6rx2W2vLiTXJPkZ3adY2wZbFDTHyY5K7O6vpXZmPCDI0XcJxv9DlbVjsz+TixVwynZ8LW6KslrMhvj/p8k/37ZThe0QV13TnLq8JRzk5yxTDPENzr+zezcdks7vthDXbfLko4v9lDTazPy+EKDEAAAAABWmCXGAAAAALDCNAgBAAAAYIVpEAIAAADACtMgBAAAAIAVpkEIAAAAACtMgxAAgLlV1faqunzsHAAALI4GIQAAo6qqbWNnAABYZRqEAADsrYOq6g+q6oqquqCq7lBVR1bVxVX1V1X17qq6a5JU1Yeq6qjh/sFVdc1w/3lV9Y6q+uMkF4xXCgAAGoQAAOytI5K8rrsflOSrSZ6e5H8keWl3PyTJJ5O8bI7v86gkJ3f3E/ZXUAAAbp0GIQAAe+tz3X3ZcP/SJPdLcpfu/vCw7+wkj53j+1zY3V/ZD/kAANgLGoQAAOytb665/+0kd9nDc2/KzWPO2+/22D8sMBMAAPtIgxAAgM36WpK/q6rHDNvPTbJrNuE1SR4x3H/GFucCAGAOrhgHAMAinJzk96vqjkk+m+Qnh/2vSnJOVT03yQfHCgcAwMaqu8fOAAAAAACMxBJjAAAAAFhhGoQAAAAAsMI0CAEAAABghWkQAgAAAMAK0yAEAAAAgBWmQQgAAAAAK0yDEAAAAABWmAYhAAAAAKwwDUIAAAAAWGEahAAAAACwwjQIAQAAAGCFaRACAAAAwArTIAQAAACAFbZt7ACLcvDBB/f27dvHjgEAAAAAB6RLL730y919yO77J9Mg3L59ey655JKxYwAAAADAAamqPr/efkuMAQAAAGCFaRACAAAAwAqbzBJjgM14844njR1hbs9+3gfmfu5/fdPy1PWC58xfFwAAAItjBiEAAAAArDAzCAHgAPDk844fO8Lc3nf8eWNHAAAAFkiDEADYL45998vGjjC39z715WNHAACA0WgQAnvtfW88duwIc3vy8987dgRgYn783NeOHWFuf/K0F44dAQCAJeAchAAAAACwwswgBGCp/Kdzjhk7wtx+/SfeP3YEAACAW2UGIQAAAACsMA1CAAAAAFhhlhjDfvTR/3bc2BHm9sM/c/7YEQAAAIARmEEIAAAAACtMgxAAAAAAVpgGIQAAAACsMA1CAAAAAFhhGoQAAAAAsMI0CAEAAABghWkQAgAAAMAK0yAEAAAAgBWmQQgAAAAAK0yDEAAAAABW2JY0CKvq8Kr606q6sqquqKrThv13q6oLq+ozw+1d13zNGVV1dVVdVVVP2oqcAAAAALBqtmoG4U1JXtLdD0jyyCSnVtUDk5ye5KLuPiLJRcN2hsdOSvKgJMck+d2qOmiLsgIAAADAytiSBmF3X9fdHx/u/32SK5McmuSEJGcPTzs7yVOG+yckeVt3f7O7P5fk6iRHb0VWAAAAAFglW34OwqranuRhST6a5B7dfV0yayImufvwtEOTfHHNl+0c9u3+vU6pqkuq6pIbb7xxv+YGAAAAgCna0gZhVd05ybuSvKi7v76np66zr2+xo/v13X1Udx91yCGHLComAAAAAKyMLWsQVtVtMmsOvrm7zx12X19V9xwev2eSG4b9O5McvubLD0ty7VZlBQAAAIBVsVVXMa4kb0xyZXe/es1D5yU5ebh/cpL3rNl/UlXdrqruk+SIJB/biqwAAAAAsEq2bdHPeXSS5yb5ZFVdNuz75SSvSHJOVT0/yReSnJgk3X1FVZ2T5FOZXQH51O7+9hZlBQAAAICVsSUNwu7+n1n/vIJJ8sQNvubMJGfut1AAAAAAwNZfxRgAAAAAOHBoEAIAAADACtMgBAAAAIAVtlUXKYFbdc1rnzJ2hLltf+EfjR0BAAAAYCE0CAEAyHHv2jF2hLmc//TnjR0BAGByLDEGAAAAgBWmQQgAAAAAK0yDEAAAAABWmAYhAAAAAKwwDUIAAAAAWGEahAAAAACwwjQIAQAAAGCFaRACAAAAwArTIAQAAACAFaZBCAAAAAArTIMQAAAAAFaYBiEAAAAArDANQgAAAABYYRqEAAAAALDCNAgBAAAAYIVpEAIAAADACtMgBAAAAIAVpkEIAAAAACts29gB2HvX/95vjB1hbvf4uV8aOwIAsKKOe+c7xo4wt/OfceLYEQCAFWYGIQAAAACsMA1CAAAAAFhhGoQAAAAAsMI0CAEAAABghWkQAgAAAMAK0yAEAAAAgBWmQQgAAAAAK0yDEAAAAABWmAYhAAAAAKwwDUIAAAAAWGEahAAAAACwwraNHQAAAJjfCe98/9gR5vKeZxwzdgQAYE5mEAIAAADACpv8DMIbf+9NY0eY2yE/95yxIwAAAACwYswgBAAAAIAVpkEIAAAAACtMgxAAAAAAVpgGIQAAAACssAP6IiVVdUyS1yQ5KMkbuvsVI0cCAAAW7Gnv+sjYEeZ27tMfNXYEAFi4A3YGYVUdlOR1SZ6c5IFJnlVVDxw3FQAAAABMy4E8g/DoJFd392eTpKreluSEJJ8aNRUAAMAcnnnuZ8eOMLe3P+2+Y0cAYETV3WNnWFdVPSPJMd3908P2c5P8cHe/YM1zTklyyrB5/yRXbVG8g5N8eYt+1laaYl1TrCmZZl1TrCmZZl1TrCmZZl1TrCmZZl1TrCmZZl1TrClR1zKZYk3JNOuaYk3JNOuaYk3JNOuaYk3J1tb1g919yO47D+QZhLXOvu/qZnb365O8fmvi3KyqLunuo7b65+5vU6xrijUl06xrijUl06xrijUl06xrijUl06xrijUl06xrijUl6lomU6wpmWZdU6wpmWZdU6wpmWZdU6wpOTDqOmDPQZhkZ5LD12wfluTakbIAAAAAwCQdyA3Cv0hyRFXdp6pum+SkJOeNnAkAAAAAJuWAXWLc3TdV1QuSfCDJQUnO6u4rRo61y5Yva94iU6xrijUl06xrijUl06xrijUl06xrijUl06xrijUl06xrijUl6lomU6wpmWZdU6wpmWZdU6wpmWZdU6wpOQDqOmAvUgIAAAAA7H8H8hJjAAAAAGA/0yAEAAAAgBWmQbiXquqYqrqqqq6uqtPHzrMIVXVWVd1QVZePnWVRqurwqvrTqrqyqq6oqtPGzrRZVXX7qvpYVX1iqOnlY2dalKo6qKr+sqrOHzvLolTVNVX1yaq6rKouGTvPolTVXarqnVX16eH/16PGzrQZVXX/4TXa9e/rVfWisXMtQlW9ePhbcXlVvbWqbj92ps2qqtOGeq5Y5tdpvffdqrpbVV1YVZ8Zbu86ZsZ9sUFdJw6v13eq6qgx8+2LDWp65fA38K+q6t1VdZcRI+6TDer69aGmy6rqgqq615gZ99aexrNV9QtV1VV18BjZNmOD1+o/V9WX1rx3HTtmxr210WtVVT8/HGddUVW/MVa+fbXBa/X2Na/TNVV12YgR99oGNR1ZVRfvGuNW1dFjZtwXG9T10Kr6yDB+/+Oq+r4xM+6tjY5/l318sYe6lnZ8sYeaRh9faBDuhao6KMnrkjw5yQOTPKuqHjhuqoXYkeSYsUMs2E1JXtLdD0jyyCSnTuC1+maSJ3T3Q5McmeSYqnrkuJEW5rQkV44dYj94fHcf2d1L9aZ1K16T5P3d/S+SPDRL/rp191XDa3Rkkkck+UaSd4+bavOq6tAkL0xyVHc/OLOLfZ00bqrNqaoHJ/l3SY7O7HfvuKo6YtxU+2xHbvm+e3qSi7r7iCQXDdvLZkduWdflSZ6W5M+2PM1i7Mgta7owyYO7+yFJ/leSM7Y61ALsyC3remV3P2T4e3h+kl/d6lCbtCPrjGer6vAkP5rkC1sdaEF2ZP1x+m/tev/q7vducabN2pHdaqqqxyc5IclDuvtBSV41Qq7N2pHd6uruZ64ZZ7wrybkj5NqMHbnl799vJHn5UNOvDtvLZkduWdcbkpze3f8ys7HgL251qE3a6Ph32ccXG9W1zOOLjWoafXyhQbh3jk5ydXd/tru/leRtmb2RLbXu/rMkXxk7xyJ193Xd/fHh/t9n1sQ4dNxUm9Mz/3vYvM3wb+mvMlRVhyX58czelDmADZ+kPjbJG5Oku7/V3V8dNdRiPTHJX3f358cOsiDbktyhqrYluWOSa0fOs1kPSHJxd3+ju29K8uEkTx050z7Z4H33hCRnD/fPTvKUrcy0COvV1d1XdvdVI0XatA1qumD4HUySi5MctuXBNmmDur6+ZvNOWbIxxh7Gs7+V5JeyZPXsMtFx+no1/VySV3T3N4fn3LDlwTZpT69VVVWSn0jy1i0NtUkb1NRJds2u+2dZwvHFBnXdPzc3my5M8vQtDbVJezj+XerxxUZ1LfP4Yg81jT6+0CDcO4cm+eKa7Z1Z8qbTKqiq7UkeluSjI0fZtJotxb0syQ1JLuzupa8pyW9nNnD/zsg5Fq2TXFBVl1bVKWOHWZD7JrkxyX+v2ZLwN1TVncYOtUAnZckG7hvp7i9lNvviC0muS/K17r5g3FSbdnmSx1bV91fVHZMcm+TwkTMt0j26+7pkNnBMcveR8zCfn0ryvrFDLEpVnVlVX0zy7CzfDMJbqKrjk3ypuz8xdpb94AXDMrSzlm3J4AZ+KMljquqjVfXhqvpXYwdasMckub67PzN2kAV4UZJXDn8rXpXlnEW9nsuTHD/cPzFLPMbY7fh3MuOLKR3X77KHmkYZX2gQ7p1aZ99Sfhq5KqrqzplN53/Rbp+ML6Xu/vYwnf+wJEcPS+6WVlUdl+SG7r507Cz7waO7++GZnZLg1Kp67NiBFmBbkocn+b3ufliSf8jyLVNYV1XdNrNB4TvGzrIIw8HiCUnuk+ReSe5UVc8ZN9XmdPeVSf5LZp/qvz/JJzJbogGjqKpfyex38M1jZ1mU7v6V7j48s5peMHaezRg+SPiVTKDRuY7fS3K/zE45c12S3xw1zWJsS3LXzJbb/WKSc4ZZd1PxrEzkQ8jMZnu+ePhb8eIMK0sm4KcyG7NfmuR7k3xr5Dz7ZGrHv7tMsa6NahpzfKFBuHd25rs/STgsSzilelVU1W0y+w/35u5etvN97NGwrPNDWf5zRz46yfFVdU1mS/afUFVvGjfSYnT3tcPtDZmdx2TpTuC8jp1Jdq6ZufrOzBqGU/DkJB/v7uvHDrIgP5Lkc919Y3f/Y2bnPPrXI2fatO5+Y3c/vLsfm9nSoCnMxNjl+qq6Z5IMt0u3vG6VVNXJSY5L8uzunuKHxW/Jki2vW8f9MvuQ5BPDOOOwJB+vqh8YNdUCdPf1w4fG30nyB5nOGOPc4ZQ6H8tsZcnSXVRmPcOpPp6W5O1jZ1mQk3PzuRTfkWn8/qW7P93dP9bdj8ismfvXY2faWxsc/y79+GKKx/Ub1TT2+EKDcO/8RZIjquo+w2yTk5KcN3Im1jF84vjGJFd296vHzrMIVXXIrisZVdUdMmsAfHrUUJvU3Wd092HdvT2z/08f7O6lnuWUJFV1p6r63l33k/xYZssWllp3/02SL1bV/YddT0zyqREjLdKUPtlPZkuLH1lVdxz+Hj4xS35BmSSpqrsPt/fO7GBrSq/ZeZkddGW4fc+IWdiDqjomyUuTHN/d3xg7z6LsdtGf47P8Y4xPdvfdu3v7MM7YmeThw3vZUtt1sD94aiYwxkjyR0mekCRV9UNJbpvky2MGWqAfSfLp7t45dpAFuTbJvx3uPyET+bBuzRjje5L8xyS/P26ivbOH49+lHl9M9Lh+3ZoOhPFFTfNDz/2nqo7N7JxpByU5q7vPHDfR5lXVW5M8LrNP6a5P8rLuXuqp4lX1b5L8eZJP5uZz2/1yL99V3v5JVT0ksxPLHpRZc/+c7v61cVMtTlU9LskvdPdxI0fZtKq6b26+Eu62JG+Zwt+KJKmqIzO7oMxtk3w2yU9299+NGmqThmVoX0xy3+7+2th5FqWqXp7kmZktUfjLJD+96+Tvy6qq/jzJ9yf5xyT/obsvGjnSPlnvfTezg+Nzktw7swbvid29VBcm2KCuryT5nSSHJPlqksu6+0kjRdxrG9R0RpLbJfnb4WkXd/fPjhJwH21Q17GZnaT/O0k+n+Rnh/OZLoVbG88OswiP6u6lajpt8Fo9LrPlxZ3kmiQ/s+scY8tgg5r+MMlZmdX1rczGhB8cKeI+2eh3sKp2ZPZ3YqkaTsmGr9VVSV6T2Rj3/yT598t2uqAN6rpzklOHp5yb5IxlmiG+0fFvZue2W9rxxR7qul2WdHyxh5pem5HHFxqEAAAAALDCLDEGAAAAgBWmQQgAAAAAK0yDEAAAAABWmAYhAAAAAKwwDUIAAAAAWGEahAAAzK2qtlfV5WPnAABgcTQIAQAYVVVtGzsDAMAq0yAEAGBvHVRVf1BVV1TVBVV1h6o6sqourqq/qqp3V9Vdk6SqPlRVRw33D66qa4b7z6uqd1TVHye5YLxSAADQIAQAYG8dkeR13f2gJF9N8vQk/yPJS7v7IUk+meRlc3yfRyU5ubufsL+CAgBw6zQIAQDYW5/r7suG+5cmuV+Su3T3h4d9Zyd57Bzf58Lu/sp+yAcAwF7QIAQAYG99c839bye5yx6ee1NuHnPefrfH/mGBmQAA2EcahAAAbNbXkvxdVT1m2H5ukl2zCa9J8ojh/jO2OBcAAHNwxTgAABbh5CS/X1V3TPLZJD857H9VknOq6rlJPjhWOAAANlbdPXYGAAAAAGAklhgDAAAAwArTIAQAAACAFaZBCAAAAAArTIMQAAAAAFaYBiEAAAAArDANQgAAAABYYRqEAAAAALDCNAgBAAAAYIVpEAIAAADACtMgBAAAAIAVpkEIAAAAACtMgxAAAAAAVpgGIQAAAACssG1jB1iUgw8+uLdv3z52DAAAAAA4IF166aVf7u5Ddt8/mQbh9u3bc8kll4wdAwAAAAAOSFX1+fX2W2IMAAAAACtMgxAAAAAAVthklhgDAMBax73zHWNHmNv5zzhx7AgAwArTIAQA9otj3/2ysSPM7b1PffnYEQAAYDQahAAAsEROeOf7x44wl/c845ixIwAAc9IgBAAgx71rx9gR5nL+0583dgQAgMlxkRIAAAAAWGEahAAAAACwwiwxBgDYCz9+7mvHjjC3P3naC8eOAHN52rs+MnaEuZ379EeNHQEAFs4MQgAAAABYYWYQAsAB4MnnHT92hLm97/jzxo4AAAAs0JbMIKyqw6vqT6vqyqq6oqpOG/bfraourKrPDLd3XfM1Z1TV1VV1VVU9aStyAgAAAMCq2aolxjcleUl3PyDJI5OcWlUPTHJ6kou6+4gkFw3bGR47KcmDkhyT5Her6qAtygoAAAAAK2NLGoTdfV13f3y4//dJrkxyaJITkpw9PO3sJE8Z7p+Q5G3d/c3u/lySq5McvRVZAQAAAGCVbPlFSqpqe5KHJfloknt093XJrImY5O7D0w5N8sU1X7Zz2Lf79zqlqi6pqktuvPHG/ZobAAAAAKZoSxuEVXXnJO9K8qLu/vqenrrOvr7Fju7Xd/dR3X3UIYccsqiYAAAAALAytqxBWFW3yaw5+ObuPnfYfX1V3XN4/J5Jbhj270xy+JovPyzJtVuVFQAAAABWxVZdxbiSvDHJld396jUPnZfk5OH+yUnes2b/SVV1u6q6T5IjknxsK7ICAAAAwCrZtkU/59FJnpvkk1V12bDvl5O8Isk5VfX8JF9IcmKSdPcVVXVOkk9ldgXkU7v721uUFQAAAABWxpY0CLv7f2b98womyRM3+Jozk5y530IBAAAAAFs2gxAAAGClPPPcz44dYW5vf9p9x44AwIi29CrGAAAAAMCBRYMQAAAAAFaYBiEAAAAArDANQgAAAABYYS5SAsBS+U/nHDN2hLn9+k+8f+wIAAAAt8oMQgAAAABYYRqEAAAAALDCNAgBAAAAYIVpEAIAAADACtMgBAAAAIAVpkEIAAAAACtMgxAAAAAAVpgGIQAAAACssE01CKvqtlV16qLCAAAAAABba64GYVU9sapeUlUnDNvbquqFST6X5Gf3Z0AAAAAAYP/ZdmtPqKqXJvlPSa5I8qCq+t0kj0vyzSSndPef7NeEAAAAAMB+c6sNwiQ/k+TfdvelVfXIJP/fJL/Q3b+1f6MBAAAAAPvbPEuMD+7uS5Okuy/ObObgb+/PUAAAAADA1phnBmGqqpLs+vd/hn3/1Fzs7u/sl3QAAAAAwH41T4PwzkluWrNda7YrSSc5aMG5AFiA//qmJ40dYW4veM4Hxo4AAACwkuZpED46ybX7OwgAAAAAsPXmOQfhB7r787v+JfmttdvDvj2qqrOq6oaqunzNvrtV1YVV9Znh9q5rHjujqq6uqquqanmmvwAAAADAkpmnQVi7bT9uH37OjiTH7Lbv9CQXdfcRSS4atlNVD0xyUpIHDV/zu1VlCTMAAAAA7AfzLDHuzf6Q7v6zqtq+2+4TcnOz8ewkH0ry0mH/27r7m0k+V1VXJzk6yUc2mwMAAIB995p3/83YEeZ22lN/YOwIAEtjngbhtqp6fG6eSbj7drr7g/vws+/R3dcNX39dVd192H9okovXPG/nsO8WquqUJKckyb3vfe99iAAAAAAAq22eBuENSc5as/23u213kvsuMNPuS5p3/Yxb7ux+fZLXJ8lRRx216ZmOAAAAALBqbrVB2N3b99PPvr6q7jnMHrxnZo3IZDZj8PA1zzssrqIMAAAAAPvFPBcp2V/OS3LycP/kJO9Zs/+kqrpdVd0nyRFJPjZCPgAAAACYvHmWGG9aVb01swuSHFxVO5O8LMkrkpxTVc9P8oUkJyZJd19RVeck+VSSm5Kc2t3f3oqcAAAAALBqtqRB2N3P2uChJ27w/DOTnLn/EgEAAAAAybhLjAEAAACAkWkQAgAAAMAK0yAEAAAAgBWmQQgAAAAAK0yDEAAAAABWmAYhAAAAAKwwDUIAAAAAWGEahAAAAACwwraNHQDgQPDmHU8aO8Lcnv28D4wdAQAAgAkxgxAAAAAAVpgGIQAAAACsMA1CAAAAAFhhzkEIAADASnvPO748doS5nXDiwWNHACbIDEIAAAAAWGEahAAAAACwwjQIAQAAAGCFaRACAAAAwArTIAQAAACAFaZBCAAAAAArTIMQAAAAAFbYtrEDAAAAAIv1kbNvHDvC3B518iFjR4CVZwYhAAAAAKywA3oGYVUdk+Q1SQ5K8obufsXIkYAk73vjsWNHmNuTn//esSMAAADAAe2AbRBW1UFJXpfkR5PsTPIXVXVed39q3GQAAADAGK7+nevHjjC3f/7z9xg7AsztQF5ifHSSq7v7s939rSRvS3LCyJkAAAAAYFIO2BmESQ5N8sU12zuT/PBIWdgC17z2KWNHmNv2F/7RXM/76H87bv8GWaAf/pnzx44AAACwcv7mNz8zdoS5/cBLjhg7AvtJdffYGdZVVScmeVJ3//Sw/dwkR3f3z695zilJThk275/kqi2Kd3CSL2/Rz9pKU6xrijUl06xrijUl06xrijUl06xrijUl06xrijUl06xrijUl6lomU6wpmWZdU6wpmWZdU6wpmWZdU6wp2dq6frC7b3Hp8AN5BuHOJIev2T4sybVrn9Ddr0/y+q0MlSRVdUl3H7XVP3d/m2JdU6wpmWZdU6wpmWZdU6wpmWZdU6wpmWZdU6wpmWZdU6wpUdcymWJNyTTrmmJNyTTrmmJNyTTrmmJNyYFR14F8DsK/SHJEVd2nqm6b5KQk542cCQAAAAAm5YCdQdjdN1XVC5J8IMlBSc7q7itGjgUAAAAAk3LANgiTpLvfm+S9Y+dYx5Yva94iU6xrijUl06xrijUl06xrijUl06xrijUl06xrijUl06xrijUl6lomU6wpmWZdU6wpmWZdU6wpmWZdU6wpOQDqOmAvUgIAAAAA7H8H8jkIAQAAAID9TINwL1XVMVV1VVVdXVWnj51nEarqrKq6oaouHzvLolTV4VX1p1V1ZVVdUVWnjZ1ps6rq9lX1sar6xFDTy8fOtChVdVBV/WVVnT92lkWpqmuq6pNVdVlVXTJ2nkWpqrtU1Tur6tPD/69HjZ1pM6rq/sNrtOvf16vqRWPnWoSqevHwt+LyqnprVd1+7EybVVWnDfVcscyv03rvu1V1t6q6sKo+M9zedcyM+2KDuk4cXq/vVNXSXXFwg5peOfwN/KuqendV3WXEiPtkg7p+fajpsqq6oKruNWbGvbWn8WxV/UJVdVUdPEa2zdjgtfrPVfWlNe9dx46ZcW9t9FpV1c8Px1lXVNVvjJVvX23wWr19zet0TVVdNmLEvbZBTUdW1cW7xrhVdfSYGffFBnU9tKo+Mozf/7iqvm/MjHtro+PfZR9f7KGupR1f7KGm0ccXGoR7oaoOSvK6JE9O8sAkz6qqB46baiF2JDlm7BALdlOSl3T3A5I8MsmpE3itvpnkCd390CRHJjmmqh45bqSFOS3JlWOH2A8e391Hjn25+gV7TZL3d/e/SPLQLPnr1t1XDa/RkUkekeQbSd49bqrNq6pDk7wwyVHd/eDMLvZ10ripNqeqHpzk3yU5OrPfveOq6ohxU+2zHbnl++7pSS7q7iOSXDRsL5sduWVdlyd5WpI/2/I0i7Ejt6zpwiQP7u6HJPlfSc7Y6lALsCO3rOuV3f2Q4e/h+Ul+datDbdKOrDOerarDk/xoki9sdaAF2ZH1x+m/tev9azhv+zLZkd1qqqrHJzkhyUO6+0FJXjVCrs3akd3q6u5nrhlnvCvJuSPk2owdueXv328keflQ068O28tmR25Z1xuSnN7d/zKzseAvbnWoTdro+HfZxxcb1bXM44uNahp9fKFBuHeOTnJ1d3+2u7+V5G2ZvZEtte7+syRfGTvHInX3dd398eH+32fWxDh03FSb0zP/e9i8zfBv6U8iWlWHJfnxzN6UOYANn6Q+Nskbk6S7v9XdXx011GI9Mclfd/fnxw6yINuS3KGqtiW5Y5JrR86zWQ9IcnF3f6O7b0ry4SRPHTnTPtngffeEJGcP989O8pStzLQI69XV3Vd291UjRdq0DWq6YPgdTJKLkxy25cE2aYO6vr5m805ZsjHGHsazv5Xkl7Jk9ewy0XH6ejX9XJJXdPc3h+fcsOXBNmlPr1VVVZKfSPLWLQ21SRvU1El2za77Z1nC8cUGdd0/NzebLkzy9C0NtUl7OP5d6vHFRnUt8/hiDzWNPr7QINw7hyb54prtnVnyptMqqKrtSR6W5KMjR9m0mi3FvSzJDUku7O6lrynJb2c2cP/OyDkWrZNcUFWXVtUpY4dZkPsmuTHJf6/ZkvA3VNWdxg61QCdlyQbuG+nuL2U2++ILSa5L8rXuvmDcVJt2eZLHVtX3V9Udkxyb5PCRMy3SPbr7umQ2cExy95HzMJ+fSvK+sUMsSlWdWVVfTPLsLN8MwluoquOTfKm7PzF2lv3gBcMytLOWbcngBn4oyWOq6qNV9eGq+ldjB1qwxyS5vrs/M3aQBXhRklcOfyteleWcRb2ey5McP9w/MUs8xtjt+Hcy44spHdfvsoeaRhlfaBDunVpn31J+GrkqqurOmU3nf9Fun4wvpe7+9jCd/7AkRw9L7pZWVR2X5IbuvnTsLPvBo7v74ZmdkuDUqnrs2IEWYFuShyf5ve5+WJJ/yPItU1hXVd02s0HhO8bOsgjDweIJSe6T5F5J7lRVzxk31eZ095VJ/ktmn+q/P8knMluiAaOoql/J7HfwzWNnWZTu/pXuPjyzml4wdp7NGD5I+JVMoNG5jt9Lcr/MTjlzXZLfHDXNYmxLctfMltv9YpJzhll3U/GsTORDyMxme754+Fvx4gwrSybgpzIbs1+a5HuTfGvkPPtkase/u0yxro1qGnN8oUG4d3bmuz9JOCxLOKV6VVTVbTL7D/fm7l62833s0bCs80NZ/nNHPjrJ8VV1TWZL9p9QVW8aN9JidPe1w+0NmZ3HZOlO4LyOnUl2rpm5+s7MGoZT8OQkH+/u68cOsiA/kuRz3X1jd/9jZuc8+tcjZ9q07n5jdz+8ux+b2dKgKczE2OX6qrpnkgy3S7e8bpVU1clJjkvy7O6e4ofFb8mSLa9bx/0y+5DkE8M447AkH6+qHxg11QJ09/XDh8bfSfIHmc4Y49zhlDofy2xlydJdVGY9w6k+npbk7WNnWZCTc/O5FN+Rafz+pbs/3d0/1t2PyKyZ+9djZ9pbGxz/Lv34YorH9RvVNPb4QoNw7/xFkiOq6j7DbJOTkpw3cibWMXzi+MYkV3b3q8fOswhVdciuKxlV1R0yawB8etRQm9TdZ3T3Yd29PbP/Tx/s7qWe5ZQkVXWnqvreXfeT/FhmyxaWWnf/TZIvVtX9h11PTPKpESMt0pQ+2U9mS4sfWVV3HP4ePjFLfkGZJKmquw+3987sYGtKr9l5mR10Zbh9z4hZ2IOqOibJS5Mc393fGDvPoux20Z/js/xjjE929927e/swztiZ5OHDe9lS23WwP3hqJjDGSPJHSZ6QJFX1Q0lum+TLYwZaoB9J8unu3jl2kAW5Nsm/He4/IRP5sG7NGON7kvzHJL8/bqK9s4fj36UeX0z0uH7dmg6E8UVN80PP/aeqjs3snGkHJTmru88cN9HmVdVbkzwus0/prk/ysu5e6qniVfVvkvx5kk/m5nPb/XIv31Xe/klVPSSzE8selFlz/5zu/rVxUy1OVT0uyS9093EjR9m0qrpvbr4S7rYkb5nC34okqaojM7ugzG2TfDbJT3b3340aapOGZWhfTHLf7v7a2HkWpapenuSZmS1R+MskP73r5O/Lqqr+PMn3J/nHJP+huy8aOdI+We99N7OD43OS3DuzBu+J3b1UFybYoK6vJPmdJIck+WqSy7r7SSNF3Gsb1HRGktsl+dvhaRd398+OEnAfbVDXsZmdpP87ST6f5GeH85kuhVsbzw6zCI/q7qVqOm3wWj0us+XFneSaJD+z6xxjy2CDmv4wyVmZ1fWtzMaEHxwp4j7Z6HewqnZk9ndiqRpOyYav1VVJXpPZGPf/JPn3y3a6oA3qunOSU4ennJvkjGWaIb7R8W9m57Zb2vHFHuq6XZZ0fLGHml6bkccXGoQAAAAAsMIsMQYAAACAFaZBCAAAAAArTIMQAAAAAFaYBiEAAAAArDANQgAAAABYYRqEAADMraq2V9XlY+cAAGBxNAgBABhVVW0bOwMAwCrTIAQAYG8dVFV/UFVXVNUFVXWHqjqyqi6uqr+qqndX1V2TpKo+VFVHDfcPrqprhvvPq6p3VNUfJ7lgvFIAANAgBABgbx2R5HXd/aAkX03y9CT/I8lLu/shST6Z5GVzfJ9HJTm5u5+wv4ICAHDrNAgBANhbn+vuy4b7lya5X5K7dPeHh31nJ3nsHN/nwu7+yn7IBwDAXtAgBABgb31zzf1vJ7nLHp57U24ec95+t8f+YYGZAADYRxqEAABs1teS/F1VPWbYfm6SXbMJr0nyiOH+M7Y4FwAAc3DFOAAAFuHkJL9fVXdM8tkkPznsf1WSc6rquUk+OFY4AAA2Vt09dgYAAAAAYCSWGAMAAADACtMgBAAAAIAVpkEIAAAAACtMgxAAAAAAVpgGIQAAAACsMA1CAAAAAFhhGoQAAAAAsMI0CAEAAABghWkQAgAAAMAK0yAEAAAAgBWmQQgAAAAAK0yDEAAAAABWmAYhAAAAAKywbWMHWJSDDz64t2/fPnYMAAAAADggXXrppV/u7kN23z+ZBuH27dtzySWXjB0DAAAAAA5IVfX59fZbYgwAAAAAK0yDEAAAAABW2GSWGAMAwCo44Z3vHzvCXN7zjGPGjgAAzMkMQgAAAABYYRqEAAAAALDCNAgBAAAAYIU5ByEAsF8c++6XjR1hbu996svHjgAAAKMxgxAAAAAAVpgGIQAAAACssC1pEFbV4VX1p1V1ZVVdUVWnDfvvVlUXVtVnhtu7rvmaM6rq6qq6qqqetBU5AQAAAGDVbNUMwpuSvKS7H5DkkUlOraoHJjk9yUXdfUSSi4btDI+dlORBSY5J8rtVddAWZQUAAACAlbElDcLuvq67Pz7c//skVyY5NMkJSc4ennZ2kqcM909I8rbu/mZ3fy7J1UmO3oqsAAAAALBKtvwqxlW1PcnDknw0yT26+7pk1kSsqrsPTzs0ycVrvmznsG/373VKklOS5N73vvd+TA0AMPPj57527Ahz+5OnvXDsCAAALIEtbRBW1Z2TvCvJi7r761W14VPX2de32NH9+iSvT5KjjjrqFo8DwLJ48nnHjx1hbu87/ryxIwAAAAu0ZVcxrqrbZNYcfHN3nzvsvr6q7jk8fs8kNwz7dyY5fM2XH5bk2q3KCgAAAACrYktmENZsquAbk1zZ3a9e89B5SU5O8orh9j1r9r+lql6d5F5Jjkjysa3ICgDANBz3zneMHWFu5z/jxLEjAAArbKuWGD86yXOTfLKqLhv2/XJmjcFzqur5Sb6Q5MQk6e4rquqcJJ/K7ArIp3b3t7coKwAAAACsjC1pEHb3/8z65xVMkidu8DVnJjlzv4UCAAAAALbuHIQAAAAAwIFHgxAAAAAAVpgGIQAAAACsMA1CAAAAAFhhW3UVYwAAgHU97V0fGTvC3M59+qPGjgAAC2cGIQAAAACsMA1CAAAAAFhhlhgDsFT+0znHjB1hbr/+E+8fOwLM7bh37Rg7wlzOf/rzxo4AADA5ZhACAAAAwArTIAQAAACAFaZBCAAAAAArTIMQAAAAAFaYBiEAAAAArDANQgAAAABYYRqEAAAAALDCto0dAAAAYIqeee5nx44wt7c/7b5jRwBgRPs0g7Cq7l9VT62q7QvOAwAAAABsoVttEFbVb1bVc9Zs/99Jrkjy+iRXVtWT92M+AAAAAGA/mmcG4VOS/Nma7f8nyQu7+5AkP5vkZfshFwAAAACwBeZpEB7S3V9Ikqp6cJLvT/LG4bE3Jfmh/ZQNAAAAANjP5mkQfq2q7jHcf0ySS7r7m8P2bZLUfkkGAAAAAOx381zF+Jwkb6uqdyd5SZJXrHnsh5P89f4IBgAAAADsf/PMIDw9yYeS/GhmFyb5/TWPHZnkvy08FQAAAACwJeZpEP5md7+8u/8/3X1mkp/a9UB3vybJMbf2DarqrKq6oaouX7PvblV1YVV9Zri965rHzqiqq6vqqqp60l7WBAAAAADMaZ4G4fN2237lbts/Osf32JFbNhJPT3JRdx+R5KJhO1X1wCQnJXnQ8DW/W1UHzfEzAAAAAIC9NE+DcPeLkNza9i10958l+cpuu09IcvZw/+wkT1mz/23d/c3u/lySq5McPUdOAAAAAGAvzdMg7L3cntc9uvu6JBlu7z7sPzTJF9c8b+ew7xaq6pSquqSqLrnxxhv3MQYAAAAArK55rmK8raoen5tnCu6+vejlv+vNSFy3Cdndr8/swik56qij9rVRCQAAAAAra54G4Q1Jzlqz/be7bd+wjz/7+qq6Z3dfV1X3XPN9diY5fM3zDkty7T7+DAAAAABgD261Qdjd2/fTzz4vyclJXjHcvmfN/rdU1auT3CvJEUk+tp8yAEzaf33T8lwI/gXP+cDYEQAAAFbSPDMIN62q3prkcUkOrqqdSV6WWWPwnKp6fpIvJDkxSbr7iqo6J8mnktyU5NTu/vZW5AQAAACAVbMlDcLuftYGDz1xg+efmeTM/ZcIAAAAAEjmu4oxAAAAADBRGoQAAAAAsMK2ZIkxAAAAy+817/6bsSPM7bSn/sDYEQCWhhmEAAAAALDCNAgBAAAAYIVpEAIAAADACtMgBAAAAIAVpkEIAAAAACtMgxAAAAAAVpgGIQAAAACsMA1CAAAAAFhhGoQAAAAAsMK2jR0A4EDw5h1PGjvC3J79vA+MHQEAAIAJ0SAEAABgpb3nHV8eO8LcTjjx4LEjABNkiTEAAAAArDANQgAAAABYYRqEAAAAALDCNAgBAAAAYIVpEAIAAADACtMgBAAAAIAVtm3sAMDyed8bjx07wtye/Pz3jh0BAAAADmhmEAIAAADACjODEAAAACbmI2ffOHaEuT3q5EPGjgAr74CeQVhVx1TVVVV1dVWdPnYeAAAAAJiaA7ZBWFUHJXldkicneWCSZ1XVA8dNBQAAAADTciAvMT46ydXd/dkkqaq3JTkhyadGTQV74aP/7bixI8zth3/m/LEjAAAA7NHVv3P92BHm9s9//h5jR4C5HcgNwkOTfHHN9s4kPzxSFgAAAADmdP1rLh47wtzucdojx44wuurusTOsq6pOTPKk7v7pYfu5SY7u7p9f85xTkpwybN4/yVVbFO/gJF/eop+1laZY1xRrSqZZ1xRrSqZZ1xRrSqZZ1xRrSqZZ1xRrSqZZ1xRrStS1TKZYUzLNuqZYUzLNuqZYUzLNuqZYU7K1df1gd9/iykAH8gzCnUkOX7N9WJJr///t3X2QJXV97/H3J7uigiaggFEW74J3JT4EETcEYyQIRldCQUSJWMZCTUJMIIKJRoi5GpKiyqjRaK7Ri0KWJIqCQkSuD1AQH26VqKAgIBARUVYIYHxKtAIi3/vH6QmT3TnDzJ6z89vu835VbZ0+PWdnPt/qmXN+/e3+dc9/QVWdDpy+kqEAklxeVetX+udua0Osa4g1wTDrGmJNMMy6hlgTDLOuIdYEw6xriDXBMOsaYk1gXX0yxJpgmHUNsSYYZl1DrAmGWdcQa4Lto67t9iYlwBeAdUn2SrIDcAxwQeNMkiRJkiRJ0qBst2cQVtU9SU4APgGsAs6sqmsbx5IkSZIkSZIGZbttEAJU1UeBj7bOsYAVn9a8QoZY1xBrgmHWNcSaYJh1DbEmGGZdQ6wJhlnXEGuCYdY1xJrAuvpkiDXBMOsaYk0wzLqGWBMMs64h1gTbQV3b7U1KJEmSJEmSJG172/M1CCVJkiRJkiRtYzYIlynJhiQ3JLkxycmt80xDkjOT3JHkmtZZpiXJnkn+Ocl1Sa5NcmLrTJNK8qAkn09yVVfTqa0zTUuSVUm+lOTC1lmmJcnNSa5OcmWSy1vnmZYkOyf5YJLru7+vp7bONIkk+3TbaO7fD5Kc1DrXNCR5ZfdecU2Ss5M8qHWmSSU5savn2j5vp4U+d5M8LMnFSb7aPe7SMuPWGFPX0d32ujdJ7+44OKamN3XvgV9Ocn6SnRtG3Cpj6vqLrqYrk1yU5FEtMy7XYuPZJK9KUkl2bZFtEmO21Z8l+da8z67DWmZcrnHbKskfdPtZ1yZ5Y6t8W2vMtvrAvO10c5IrG0ZctjE17ZfksrkxbpIDWmbcGmPqelKSz3bj948k+emWGZdr3P5v38cXi9TV2/HFIjU1H1/YIFyGJKuAdwDPAR4PvDDJ49ummoqNwIbWIabsHuCPqupxwIHA8QPYVncBh1TVk4D9gA1JDmwbaWpOBK5rHWIbeEZV7df6dvVT9jbg41X1c8CT6Pl2q6obum20H/AU4EfA+W1TTS7JHsArgPVV9URGN/s6pm2qySR5IvA7wAGMfvcOT7KubaqttpEtP3dPBi6pqnXAJd3zvtnIlnVdAxwFfHrF00zHRras6WLgiVW1L/AvwCkrHWoKNrJlXW+qqn2798MLgdetdKgJbWSB8WySPYFfBb650oGmZCMLj9PfOvf51V23vU82sllNSZ4BHAnsW1VPAN7cINekNrJZXVX1gnnjjA8B5zXINYmNbPn790bg1K6m13XP+2YjW9b1HuDkqvp5RmPBV690qAmN2//t+/hiXF19Hl+Mq6n5+MIG4fIcANxYVTdV1d3A+xl9kPVaVX0a+E7rHNNUVbdV1Re75X9n1MTYo22qydTIf3RPH9D96/1FRJOsAX6N0YeytmPdkdSDgDMAquruqvpe01DTdSjwtar6RusgU7IaeHCS1cCOwK2N80zqccBlVfWjqroH+BTw3MaZtsqYz90jgbO65bOAX1/JTNOwUF1VdV1V3dAo0sTG1HRR9zsIcBmwZsWDTWhMXT+Y93QnejbGWGQ8+1bgj+lZPXMGOk5fqKbfA95QVXd1r7ljxYNNaLFtlSTAbwBnr2ioCY2pqYC5s+t+hh6OL8bUtQ/3NZsuBp63oqEmtMj+b6/HF+Pq6vP4YpGamo8vbBAuzx7ALfOeb6LnTadZkGQt8GTgc42jTCyjqbhXAncAF1dV72sC/prRwP3exjmmrYCLklyR5LjWYaZkb+BO4O8ymhL+niQ7tQ41RcfQs4H7OFX1LUZnX3wTuA34flVd1DbVxK4BDkry8CQ7AocBezbONE2PqKrbYDRwBHZvnEdL8zLgY61DTEuS05LcAryI/p1BuIUkRwDfqqqrWmfZBk7opqGd2bcpg2M8Fnh6ks8l+VSSX2gdaMqeDtxeVV9tHWQKTgLe1L1XvJl+nkW9kGuAI7rlo+nxGGOz/d/BjC+GtF8/Z5GamowvbBAuTxZY18ujkbMiyUMYnc5/0mZHxnupqn7Snc6/Bjigm3LXW0kOB+6oqitaZ9kGnlZV+zO6JMHxSQ5qHWgKVgP7A++sqicDP6R/0xQWlGQHRoPCc1tnmYZuZ/FIYC/gUcBOSX6zbarJVNV1wF8yOqr/ceAqRlM0pCaSvJbR7+B7W2eZlqp6bVXtyaimE1rnmUR3IOG1DKDRuYB3Ao9hdMmZ24C/appmOlYDuzCabvdq4JzurLuheCEDOQjJ6GzPV3bvFa+km1kyAC9jNGa/AngocHfjPFtlaPu/c4ZY17iaWo4vbBAuzyb++5GENfTwlOpZkeQBjP7g3ltVfbvex6K6aZ2fpP/XjnwacESSmxlN2T8kyT+2jTQdVXVr93gHo+uY9O4CzgvYBGyad+bqBxk1DIfgOcAXq+r21kGm5JnA16vqzqr6MaNrHv1S40wTq6ozqmr/qjqI0dSgIZyJMef2JI8E6B57N71uliQ5FjgceFFVDfFg8fvo2fS6BTyG0UGSq7pxxhrgi0l+tmmqKaiq27uDxvcC72Y4Y4zzukvqfJ7RzJLe3VRmId2lPo4CPtA6y5Qcy33XUjyXYfz+UVXXV9WzquopjJq5X2udabnG7P/2fnwxxP36cTW1Hl/YIFyeLwDrkuzVnW1yDHBB40xaQHfE8Qzguqp6S+s805Bkt7k7GSV5MKMGwPVNQ02oqk6pqjVVtZbR39OlVdXrs5wAkuyU5KFzy8CzGE1b6LWq+lfgliT7dKsOBb7SMNI0DenIPoymFh+YZMfu/fBQen5DGYAku3ePj2a0szWkbXYBo50uuscPN8yiRSTZALwGOKKqftQ6z7RsdtOfI+j/GOPqqtq9qtZ244xNwP7dZ1mvze3sd57LAMYYwD8BhwAkeSywA/DtloGm6JnA9VW1qXWQKbkV+JVu+RAGcrBu3hjjp4A/Bd7VNtHyLLL/2+vxxUD36xesaXsYX2SYBz23nSSHMbpm2irgzKo6rW2iySU5GziY0VG624HXV1WvTxVP8svAZ4Crue/adn9S/bvL239Jsi+jC8uuYtTcP6eq/rxtqulJcjDwqqo6vHGUiSXZm/vuhLsaeN8Q3isAkuzH6IYyOwA3AS+tqu82DTWhbhraLcDeVfX91nmmJcmpwAsYTVH4EvDbcxd/76sknwEeDvwY+MOquqRxpK2y0Ocuo53jc4BHM2rwHl1VvboxwZi6vgP8DbAb8D3gyqp6dqOIyzamplOABwL/1r3ssqp6eZOAW2lMXYcxukj/vcA3gJd31zPthfsbz3ZnEa6vql41ncZsq4MZTS8u4Gbgd+euMdYHY2r6B+BMRnXdzWhMeGmjiFtl3O9gko2M3id61XCCsdvqBuBtjMa4/wn8ft8uFzSmrocAx3cvOQ84pU9niI/b/2V0bbveji8WqeuB9HR8sUhNb6fx+MIGoSRJkiRJkjTDnGIsSZIkSZIkzTAbhJIkSZIkSdIMs0EoSZIkSZIkzTAbhJIkSZIkSdIMs0EoSZIkSZIkzTAbhJIkSVqyJGuTXNM6hyRJkqbHBqEkSZKaSrK6dQZJkqRZZoNQkiRJy7UqybuTXJvkoiQPTrJfksuSfDnJ+Ul2AUjyySTru+Vdk9zcLb8kyblJPgJc1K4USZIk2SCUJEnScq0D3lFVTwC+BzwP+HvgNVW1L3A18PolfJ+nAsdW1SHbKqgkSZLunw1CSZIkLdfXq+rKbvkK4DHAzlX1qW7dWcBBS/g+F1fVd7ZBPkmSJC2DDUJJkiQt113zln8C7LzIa+/hvjHngzb72g+nmEmSJElbyQahJEmSJvV94LtJnt49fzEwdzbhzcBTuuXnr3AuSZIkLYF3jJMkSdI0HAu8K8mOwE3AS7v1bwbOSfJi4NJW4SRJkjReqqp1BkmSJEmSJEmNOMVYkiRJkiRJmmE2CCVJkiRJkqQZZoNQkiRJkiRJmmE2CCVJkiRJkqQZZoNQkiRJkiRJmmE2CCVJkiRJkqQZZoNQkiRJkiRJmmE2CCVJkiRJkqQZZoNQkiRJkiRJmmE2CCVJkiRJkqQZZoNQkiRJkiRJmmE2CCVJkiRJkqQZZoNQkiRJkiRJmmGrWweYll133bXWrl3bOoYkSZIkSZK0Xbriiiu+XVW7bb5+MA3CtWvXcvnll7eOIUmSJEmSJG2XknxjofUrMsU4yZ5J/jnJdUmuTXJit/5hSS5O8tXucZd5/+eUJDcmuSHJs1cipyRJkiRJkjRrVuoahPcAf1RVjwMOBI5P8njgZOCSqloHXNI9p/vaMcATgA3A3yZZtUJZJUmSJEmSpJmxIlOMq+o24LZu+d+TXAfsARwJHNy97Czgk8BruvXvr6q7gK8nuRE4APjsSuSVJEmTO+z817eOsGQffe6prSNIkiRJzaz4XYyTrAWeDHwOeETXPJxrIu7evWwP4JZ5/21Tt27z73VcksuTXH7nnXdu09ySJEmSJEnSEK3oTUqSPAT4EHBSVf0gydiXLrCutlhRdTpwOsD69eu3+LokSZI0NEd+8OOtIyzJh5+/oXUESZK0RCt2BmGSBzBqDr63qs7rVt+e5JHd1x8J3NGt3wTsOe+/rwFuXamskiRJkiRJ0qxYqbsYBzgDuK6q3jLvSxcAx3bLxwIfnrf+mCQPTLIXsA74/EpklSRJkiRJkmbJSk0xfhrwYuDqJFd26/4EeANwTpLfAr4JHA1QVdcmOQf4CqM7IB9fVT9ZoaySJEmSJEnSzFipuxj/Pxa+riDAoWP+z2nAadsslCRJkiRJkqSVv4uxJEmSJEmSpO3Hit7FWJIkSdunwz+0sXWEJbnweS9pHUGSJGlwbBBKkiRpkA7/4LmtIyzZhc8/unUESZI0w2wQSpK0HXjOBUe0jrBkHzvigtYRJEmSJE2R1yCUJEmSJEmSZpgNQkmSJEmSJGmG2SCUJEmSJEmSZpjXIJQkSZLU1FEf+mzrCEt23vOe2jqCJElTZ4NQkiRpGX7tvLe3jrBk//eoV7SOIEmSpB5wirEkSZIkSZI0w2wQSpIkSZIkSTPMBqEkSZIkSZI0w2wQSpIkSZIkSTPMBqEkSZIkSZI0w7yLsSRJkiRtAy8476bWEZbsA0ft3TqCJKkhzyCUJEmSJEmSZphnEEqSeuV/nbOhdYQl+4vf+HjrCJIkSZJ0vzyDUJIkSZIkSZphNgglSZIkSZKkGWaDUJIkSZIkSZphNgglSZIkSZKkGWaDUJIkSZIkSZphNgglSZIkSZKkGWaDUJIkSZIkSZphNgglSZIkSZKkGWaDUJIkSZIkSZphNgglSZIkSZKkGXa/DcIkV69EEEmSJEmSJEkrbylnEK6d9IckOTPJHUmumbfuYUkuTvLV7nGXeV87JcmNSW5I8uxJf74kSZIkSZKkhS2lQVhT+DkbgQ2brTsZuKSq1gGXdM9J8njgGOAJ3f/52ySrppBBkiRJkiRJ0mZWL+E1Oyb59GIvqKqD7ufrn06ydrPVRwIHd8tnAZ8EXtOtf39V3QV8PcmNwAHAZ5eQVZIkSZIkSdIyLKVB+GPgjG3wsx9RVbcBVNVtSXbv1u8BXDbvdZu6dVtIchxwHMCjH/3obRBRkiRJkiRJGrYlNQir6qxtnuQ+WWDdgtOcq+p04HSA9evXT2MqtCRJkiRJkjRTltIgXKhhNw23J3lkd/bgI4E7uvWbgD3nvW4NcOs2yiBJkiRJWqK3nf+vrSMs2YnP/dnWESSpN5bSIHxjkkMWe0FVXboVP/sC4FjgDd3jh+etf1+StwCPAtYBn9+K7y9JM+9//2N/bgR/wm9+onUESZIkSZpJS2kQvqz7N04Bey/2DZKczeiGJLsm2QS8nlFj8JwkvwV8EzgaoKquTXIO8BXgHuD4qvrJEnJKkiRJkiRJWqb7bRBW1V6T/pCqeuGYLx065vWnAadN+nMlSZIkSZIkLe6nWgeQJEmSJEmS1I4NQkmSJEmSJGmG2SCUJEmSJEmSZpgNQkmSJEmSJGmG2SCUJEmSJEmSZpgNQkmSJEmSJGmG2SCUJEmSJEmSZpgNQkmSJEmSJGmG2SCUJEmSJEmSZpgNQkmSJEmSJGmG2SCUJEmSJEmSZpgNQkmSJEmSJGmG2SCUJEmSJEmSZpgNQkmSJEmSJGmG2SCUJEmSJEmSZpgNQkmSJEmSJGmGrW4dQJIkSZKklj587rdbR1iyI4/etXUESQPkGYSSJEmSJEnSDPMMQkkC3rvx2a0jLNmLXvKJ1hEkSZIkSQPiGYSSJEmSJEnSDLNBKEmSJEmSJM0wG4SSJEmSJEnSDPMahJKW7WNnHNY6wpI957c+2jqCJEmStOI+e9adrSMs2VOP3a11BGnmeQahJEmSJEmSNMNsEEqSJEmSJEkzzAahJEmSJEmSNMNsEEqSJEmSJEkzzAahJEmSJEmSNMO26wZhkg1JbkhyY5KTW+eRJEmSJEmShmZ16wDjJFkFvAP4VWAT8IUkF1TVV9omk5buc//n8NYRluwXf/fC1hEkSZIkaVE3/s3trSMs2f/8g0e0jiAt2XbbIAQOAG6sqpsAkrwfOBKwQThQN7/911tHWLK1r/in1hEkSZIkSZKmIlXVOsOCkjwf2FBVv909fzHwi1V1wrzXHAcc1z3dB7hhheLtCnx7hX7WShpiXUOsCYZZ1xBrgmHWNcSaYJh1DbEmGGZdQ6wJhlnXEGsC6+qTIdYEw6xriDXBMOsaYk0wzLqGWBOsbF3/o6p223zl9nwGYRZY99+6mVV1OnD6ysS5T5LLq2r9Sv/cbW2IdQ2xJhhmXUOsCYZZ1xBrgmHWNcSaYJh1DbEmGGZdQ6wJrKtPhlgTDLOuIdYEw6xriDXBMOsaYk2wfdS1Pd+kZBOw57zna4BbG2WRJEmSJEmSBml7bhB+AViXZK8kOwDHABc0ziRJkiRJkiQNynY7xbiq7klyAvAJYBVwZlVd2zjWnBWf1rxChljXEGuCYdY1xJpgmHUNsSYYZl1DrAmGWdcQa4Jh1jXEmsC6+mSINcEw6xpiTTDMuoZYEwyzriHWBNtBXdvtTUokSZIkSZIkbXvb8xRjSZIkSZIkSduYDUJJkiRJkiRphtkgXKYkG5LckOTGJCe3zjMNSc5MckeSa1pnmZYkeyb55yTXJbk2yYmtM00qyYOSfD7JVV1Np7bONC1JViX5UpILW2eZliQ3J7k6yZVJLm+dZ1qS7Jzkg0mu7/6+nto60ySS7NNto7l/P0hyUutc05Dkld17xTVJzk7yoNaZJpXkxK6ea/u8nRb63E3ysCQXJ/lq97hLy4xbY0xdR3fb694k61vm2xpjanpT9x745STnJ9m5YcStMqauv+hqujLJRUke1TLjci02nk3yqiSVZNcW2SYxZlv9WZJvzfvsOqxlxuUat62S/EG3n3Vtkje2yre1xmyrD8zbTjcnubJhxGUbU9N+SS6bG+MmOaBlxq0xpq4nJflsN37/SJKfbplxucbt//Z9fLFIXb0dXyxSU/PxhQ3CZUiyCngH8Bzg8cALkzy+baqp2AhsaB1iyu4B/qiqHgccCBw/gG11F3BIVT0J2A/YkOTAtpGm5kTgutYhtoFnVNV+VdWrD6378Tbg41X1c8CT6Pl2q6obum20H/AU4EfA+W1TTS7JHsArgPVV9URGN/s6pm2qySR5IvA7wAGMfvcOT7KubaqttpEtP3dPBi6pqnXAJd3zvtnIlnVdAxwFfHrF00zHRras6WLgiVW1L/AvwCkrHWoKNrJlXW+qqn2798MLgdetdKgJbWSB8WySPYFfBb650oGmZCMLj9PfOvf5VVUfXeFMk9rIZjUleQZwJLBvVT0BeHODXJPayGZ1VdUL5o0zPgSc1yDXJDay5e/fG4FTu5pe1z3vm41sWdd7gJOr6ucZjQVfvdKhJjRu/7fv44txdfV5fDGupubjCxuEy3MAcGNV3VRVdwPvZ/RB1mtV9WngO61zTFNV3VZVX+yW/51RE2OPtqkmUyP/0T19QPev93cZSrIG+DVGH8rajnVHUg8CzgCoqrur6ntNQ03XocDXquobrYNMyWrgwUlWAzsCtzbOM6nHAZdV1Y+q6h7gU8BzG2faKmM+d48EzuqWzwJ+fSUzTcNCdVXVdVV1Q6NIExtT00Xd7yDAZcCaFQ82oTF1/WDe053o2RhjkfHsW4E/pmf1zBnoOH2hmn4PeENV3dW95o4VDzahxbZVkgC/AZy9oqEmNKamAubOrvsZeji+GFPXPtzXbLoYeN6KhprQIvu/vR5fjKurz+OLRWpqPr6wQbg8ewC3zHu+iZ43nWZBkrXAk4HPNY4ysYym4l4J3AFcXFW9rwn4a0YD93sb55i2Ai5KckWS41qHmZK9gTuBv8toSvh7kuzUOtQUHUPPBu7jVNW3GJ198U3gNuD7VXVR21QTuwY4KMnDk+wIHAbs2TjTND2iqm6D0cAR2L1xHi3Ny4CPtQ4xLUlOS3IL8CL6dwbhFpIcAXyrqq5qnWUbOKGbhnZm36YMjvFY4OlJPpfkU0l+oXWgKXs6cHtVfbV1kCk4CXhT917xZvp5FvVCrgGO6JaPpsdjjM32fwczvhjSfv2cRWpqMr6wQbg8WWBdL49GzookD2F0Ov9Jmx0Z76Wq+kl3Ov8a4IBuyl1vJTkcuKOqrmidZRt4WlXtz+iSBMcnOah1oClYDewPvLOqngz8kP5NU1hQkh0YDQrPbZ1lGrqdxSOBvYBHATsl+c22qSZTVdcBf8noqP7HgasYTdGQmkjyWka/g+9tnWVaquq1VbUno5pOaJ1nEt2BhNcygEbnAt4JPIbRJWduA/6qaZrpWA3swmi63auBc7qz7obihQzkICSjsz1f2b1XvJJuZskAvIzRmP0K4KHA3Y3zbJWh7f/OGWJd42pqOb6wQbg8m/jvRxLW0MNTqmdFkgcw+oN7b1X17Xofi+qmdX6S/l878mnAEUluZjRl/5Ak/9g20nRU1a3d4x2MrmPSuws4L2ATsGnemasfZNQwHILnAF+sqttbB5mSZwJfr6o7q+rHjK559EuNM02sqs6oqv2r6iBGU4OGcCbGnNuTPBKge+zd9LpZkuRY4HDgRVU1xIPF76Nn0+sW8BhGB0mu6sYZa4AvJvnZpqmmoKpu7w4a3wu8m+GMMc7rLqnzeUYzS3p3U5mFdJf6OAr4QOssU3Is911L8VyG8ftHVV1fVc+qqqcwauZ+rXWm5Rqz/9v78cUQ9+vH1dR6fGGDcHm+AKxLsld3tskxwAWNM2kB3RHHM4DrquotrfNMQ5Ld5u5klOTBjBoA1zcNNaGqOqWq1lTVWkZ/T5dWVa/PcgJIslOSh84tA89iNG2h16rqX4FbkuzTrToU+ErDSNM0pCP7MJpafGCSHbv3w0Pp+Q1lAJLs3j0+mtHO1pC22QWMdrroHj/cMIsWkWQD8BrgiKr6Ues807LZTX+OoP9jjKuraveqWtuNMzYB+3efZb02t7PfeS4DGGMA/wQcApDkscAOwLdbBpqiZwLXV9Wm1kGm5FbgV7rlQxjIwbp5Y4yfAv4UeFfbRMuzyP5vr8cXA92vX7Cm7WF8kWEe9Nx2khzG6Jppq4Azq+q0tokml+Rs4GBGR+luB15fVb0+VTzJLwOfAa7mvmvb/Un17y5v/yXJvowuLLuKUXP/nKr687appifJwcCrqurwxlEmlmRv7rsT7mrgfUN4rwBIsh+jG8rsANwEvLSqvts01IS6aWi3AHtX1fdb55mWJKcCL2A0ReFLwG/PXfy9r5J8Bng48GPgD6vqksaRtspCn7uMdo7PAR7NqMF7dFX16sYEY+r6DvA3wG7A94Arq+rZjSIu25iaTgEeCPxb97LLqurlTQJupTF1HcboIv33At8AXt5dz7QX7m88251FuL6qetV0GrOtDmY0vbiAm4HfnbvGWB+MqekfgDMZ1XU3ozHhpY0ibpVxv4NJNjJ6n+hVwwnGbqsbgLcxGuP+J/D7fbtc0Ji6HgIc373kPOCUPp0hPm7/l9G17Xo7vlikrgfS0/HFIjW9ncbjCxuEkiRJkiRJ0gxzirEkSZIkSZI0w2wQSpIkSZIkSTPMBqEkSZIkSZI0w2wQSpIkSZIkSTPMBqEkSZIkSZI0w2wQSpIkacmSrE1yTesckiRJmh4bhJIkSWoqyerWGSRJkmaZDUJJkiQt16ok705ybZKLkjw4yX5JLkvy5STnJ9kFIMknk6zvlndNcnO3/JIk5yb5CHBRu1IkSZJkg1CSJEnLtQ54R1U9Afge8Dzg74HXVNW+wNXA65fwfZ4KHFtVh2yroJIkSbp/NgglSZK0XF+vqiu75SuAxwA7V9WnunVnAQct4ftcXFXf2Qb5JEmStAw2CCVJkrRcd81b/gmw8yKvvYf7xpwP2uxrP5xiJkmSJG0lG4SSJEma1PeB7yZ5evf8xcDc2YQ3A0/plp+/wrkkSZK0BN4xTpIkSdNwLPCuJDsCNwEv7da/GTgnyYuBS1uFkyRJ0nipqtYZJEmSJEmSJDXiFGNJkiRJkiRphtkglCRJkiRJkmaYDUJJkiRJkiRphtkglCRJkiRJkmaYDUJJkiRJkiRphtkglCRJkiRJkmaYDUJJkiRJkiRphv1/AytJcceMfHgAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 1296x864 with 7 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# visualization\n",
    "country_list = ['US', 'DE', 'UK', '', 'FR', 'ES', 'IT']\n",
    "order = list(range(0, 23))\n",
    "fig, ax = plt.subplots(nrows=7, ncols=1, figsize=(18, 12))\n",
    "for i in range(len(country_list)):\n",
    "    sns.countplot(x='hour', data=data[data['user_country'] == country_list[i]], ax=ax[i], order=order)\n",
    "    ax[i].set_ylabel(country_list[i], fontsize=12)\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "894fbe88",
   "metadata": {},
   "source": [
    "Based on above charts, we can see that, for the unknown country, their peak search time has around 12 hours difference with United States, and around 5 hours difference with Europe. Based on the geographic locations of each country, it seems that this missing country is in Asia. Possible countries include China, Japan, and South Korea."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "343b456f",
   "metadata": {},
   "source": [
    "### Part 3:For each city, find the most likely city to be also searched for within the same session."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7cde6f03",
   "metadata": {},
   "source": [
    "I want to create a matrix that will display the city-city similarity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "b24d8d53",
   "metadata": {},
   "outputs": [],
   "source": [
    "from collections import Counter\n",
    "\n",
    "data['cities'] = data.cities.str.split(', ')\n",
    "def count_cities(df):\n",
    "    c = Counter(city for cities in df.cities for city in cities)\n",
    "    return pd.Series(c)\n",
    "searchcity_by_user = data.groupby(\"user_id\").apply(count_cities).unstack(fill_value=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "4f70abcd",
   "metadata": {},
   "outputs": [],
   "source": [
    "searchcity_by_user = searchcity_by_user.transpose()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "4780e0ac",
   "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>user_id</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>9</th>\n",
       "      <th>10</th>\n",
       "      <th>11</th>\n",
       "      <th>13</th>\n",
       "      <th>15</th>\n",
       "      <th>17</th>\n",
       "      <th>18</th>\n",
       "      <th>20</th>\n",
       "      <th>...</th>\n",
       "      <th>10972</th>\n",
       "      <th>10973</th>\n",
       "      <th>10977</th>\n",
       "      <th>10979</th>\n",
       "      <th>10980</th>\n",
       "      <th>10981</th>\n",
       "      <th>10982</th>\n",
       "      <th>10988</th>\n",
       "      <th>10990</th>\n",
       "      <th>11000</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Anaheim CA</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Arlington TX</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Atlanta GA</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Austin TX</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Bakersfield CA</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 5777 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "user_id         4      5      9      10     11     13     15     17     18     \\\n",
       "Anaheim CA          0      0      0      0      0      0      0      0      0   \n",
       "Arlington TX        0      0      0      0      0      0      0      0      0   \n",
       "Atlanta GA          0      0      0      0      0      0      0      0      0   \n",
       "Austin TX           0      0      0      0      0      0      0      0      0   \n",
       "Bakersfield CA      0      0      0      0      0      0      0      0      0   \n",
       "\n",
       "user_id         20     ...  10972  10973  10977  10979  10980  10981  10982  \\\n",
       "Anaheim CA          0  ...      0      1      0      0      0      0      0   \n",
       "Arlington TX        0  ...      0      0      0      0      0      0      0   \n",
       "Atlanta GA          0  ...      0      0      0      0      0      0      0   \n",
       "Austin TX           0  ...      0      0      0      0      0      0      0   \n",
       "Bakersfield CA      0  ...      0      0      0      0      0      0      0   \n",
       "\n",
       "user_id         10988  10990  11000  \n",
       "Anaheim CA          0      0      0  \n",
       "Arlington TX        0      0      0  \n",
       "Atlanta GA          0      0      0  \n",
       "Austin TX           0      0      0  \n",
       "Bakersfield CA      0      0      0  \n",
       "\n",
       "[5 rows x 5777 columns]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "searchcity_by_user.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "88217859",
   "metadata": {},
   "outputs": [],
   "source": [
    "# make each row unit-norm, then dot-product is equivalent to cosine\n",
    "from sklearn.preprocessing import normalize\n",
    "\n",
    "searchcity_by_user_normed = normalize(searchcity_by_user,axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "d1fddc1b",
   "metadata": {},
   "outputs": [],
   "source": [
    "city_similarity = searchcity_by_user_normed.dot(searchcity_by_user_normed.T)\n",
    "city_similarity = pd.DataFrame(city_similarity,\n",
    "                               index = searchcity_by_user.index,\n",
    "                               columns = searchcity_by_user.index)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "d2174a76",
   "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>Anaheim CA</th>\n",
       "      <th>Arlington TX</th>\n",
       "      <th>Atlanta GA</th>\n",
       "      <th>Austin TX</th>\n",
       "      <th>Bakersfield CA</th>\n",
       "      <th>Baltimore MD</th>\n",
       "      <th>Birmingham AL</th>\n",
       "      <th>Boston MA</th>\n",
       "      <th>Buffalo NY</th>\n",
       "      <th>Calgary AB</th>\n",
       "      <th>...</th>\n",
       "      <th>Toledo OH</th>\n",
       "      <th>Toronto ON</th>\n",
       "      <th>Tucson AZ</th>\n",
       "      <th>Tulsa OK</th>\n",
       "      <th>Vancouver BC</th>\n",
       "      <th>Victoria BC</th>\n",
       "      <th>Virginia Beach VA</th>\n",
       "      <th>WASHINGTON DC</th>\n",
       "      <th>Wichita KS</th>\n",
       "      <th>Windsor ON</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Anaheim CA</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.033731</td>\n",
       "      <td>0.026082</td>\n",
       "      <td>0.037796</td>\n",
       "      <td>0.019139</td>\n",
       "      <td>0.038984</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.047099</td>\n",
       "      <td>0.029180</td>\n",
       "      <td>0.077773</td>\n",
       "      <td>...</td>\n",
       "      <td>0.022195</td>\n",
       "      <td>0.118564</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.018019</td>\n",
       "      <td>0.092873</td>\n",
       "      <td>0.037937</td>\n",
       "      <td>0.014494</td>\n",
       "      <td>0.026239</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.013199</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Arlington TX</th>\n",
       "      <td>0.033731</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.243935</td>\n",
       "      <td>0.015064</td>\n",
       "      <td>0.036820</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.037070</td>\n",
       "      <td>0.007655</td>\n",
       "      <td>0.040808</td>\n",
       "      <td>...</td>\n",
       "      <td>0.017469</td>\n",
       "      <td>0.058240</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.113455</td>\n",
       "      <td>0.061675</td>\n",
       "      <td>0.039812</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.020652</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.020777</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Atlanta GA</th>\n",
       "      <td>0.026082</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.013801</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.028470</td>\n",
       "      <td>0.494872</td>\n",
       "      <td>0.017198</td>\n",
       "      <td>0.017758</td>\n",
       "      <td>0.025817</td>\n",
       "      <td>...</td>\n",
       "      <td>0.020261</td>\n",
       "      <td>0.030704</td>\n",
       "      <td>0.051434</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.037092</td>\n",
       "      <td>0.015392</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.023953</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Austin TX</th>\n",
       "      <td>0.037796</td>\n",
       "      <td>0.243935</td>\n",
       "      <td>0.013801</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.045383</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.039876</td>\n",
       "      <td>0.020587</td>\n",
       "      <td>0.062354</td>\n",
       "      <td>...</td>\n",
       "      <td>0.023489</td>\n",
       "      <td>0.132595</td>\n",
       "      <td>0.014907</td>\n",
       "      <td>0.019069</td>\n",
       "      <td>0.096751</td>\n",
       "      <td>0.049071</td>\n",
       "      <td>0.015339</td>\n",
       "      <td>0.013884</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.027937</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Bakersfield CA</th>\n",
       "      <td>0.019139</td>\n",
       "      <td>0.015064</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.010446</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.037860</td>\n",
       "      <td>0.013031</td>\n",
       "      <td>0.012630</td>\n",
       "      <td>...</td>\n",
       "      <td>0.044603</td>\n",
       "      <td>0.038303</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.048280</td>\n",
       "      <td>0.038883</td>\n",
       "      <td>0.022589</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.035366</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 89 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                Anaheim CA  Arlington TX  Atlanta GA  Austin TX  \\\n",
       "Anaheim CA        1.000000      0.033731    0.026082   0.037796   \n",
       "Arlington TX      0.033731      1.000000    0.000000   0.243935   \n",
       "Atlanta GA        0.026082      0.000000    1.000000   0.013801   \n",
       "Austin TX         0.037796      0.243935    0.013801   1.000000   \n",
       "Bakersfield CA    0.019139      0.015064    0.000000   0.000000   \n",
       "\n",
       "                Bakersfield CA  Baltimore MD  Birmingham AL  Boston MA  \\\n",
       "Anaheim CA            0.019139      0.038984       0.000000   0.047099   \n",
       "Arlington TX          0.015064      0.036820       0.000000   0.037070   \n",
       "Atlanta GA            0.000000      0.028470       0.494872   0.017198   \n",
       "Austin TX             0.000000      0.045383       0.000000   0.039876   \n",
       "Bakersfield CA        1.000000      0.010446       0.000000   0.037860   \n",
       "\n",
       "                Buffalo NY  Calgary AB  ...  Toledo OH  Toronto ON  Tucson AZ  \\\n",
       "Anaheim CA        0.029180    0.077773  ...   0.022195    0.118564   0.000000   \n",
       "Arlington TX      0.007655    0.040808  ...   0.017469    0.058240   0.000000   \n",
       "Atlanta GA        0.017758    0.025817  ...   0.020261    0.030704   0.051434   \n",
       "Austin TX         0.020587    0.062354  ...   0.023489    0.132595   0.014907   \n",
       "Bakersfield CA    0.013031    0.012630  ...   0.044603    0.038303   0.000000   \n",
       "\n",
       "                Tulsa OK  Vancouver BC  Victoria BC  Virginia Beach VA  \\\n",
       "Anaheim CA      0.018019      0.092873     0.037937           0.014494   \n",
       "Arlington TX    0.113455      0.061675     0.039812           0.000000   \n",
       "Atlanta GA      0.000000      0.037092     0.015392           0.000000   \n",
       "Austin TX       0.019069      0.096751     0.049071           0.015339   \n",
       "Bakersfield CA  0.048280      0.038883     0.022589           0.000000   \n",
       "\n",
       "                WASHINGTON DC  Wichita KS  Windsor ON  \n",
       "Anaheim CA           0.026239         0.0    0.013199  \n",
       "Arlington TX         0.020652         0.0    0.020777  \n",
       "Atlanta GA           0.023953         0.0    0.000000  \n",
       "Austin TX            0.013884         0.0    0.027937  \n",
       "Bakersfield CA       0.000000         0.0    0.035366  \n",
       "\n",
       "[5 rows x 89 columns]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "city_similarity.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "c8260d59",
   "metadata": {},
   "outputs": [],
   "source": [
    "### find top K most similar of each city\n",
    "\n",
    "def most_similar(s,topk):\n",
    "    # [0] must be itself\n",
    "    similar_ones = s.sort_values(ascending=False)[1:topk+1].index.values\n",
    "    return pd.Series(similar_ones,index = [\"similar#{}\".format(i) for i in range(1,topk+1)])\n",
    "\n",
    "most_similar_cities = city_similarity.apply(most_similar,topk=1,axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "c3e87550",
   "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>similar#1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Jersey City NJ</th>\n",
       "      <td>Newark NJ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Columbus OH</th>\n",
       "      <td>Cincinnati OH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Victoria BC</th>\n",
       "      <td>Seattle WA</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Buffalo NY</th>\n",
       "      <td>Saint Catharines-Niagara ON</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Calgary AB</th>\n",
       "      <td>Seattle WA</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Norfolk VA</th>\n",
       "      <td>Virginia Beach VA</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Mesa AZ</th>\n",
       "      <td>Chandler AZ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Fort Worth TX</th>\n",
       "      <td>Arlington TX</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Arlington TX</th>\n",
       "      <td>Fort Worth TX</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>San Francisco CA</th>\n",
       "      <td>Oakland CA</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Atlanta GA</th>\n",
       "      <td>Birmingham AL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Scottsdale AZ</th>\n",
       "      <td>Mesa AZ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Minneapolis MN</th>\n",
       "      <td>Saint Paul MN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Oklahoma City OK</th>\n",
       "      <td>Tulsa OK</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Chicago IL</th>\n",
       "      <td>New York NY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Lexington KY</th>\n",
       "      <td>Louisville KY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Austin TX</th>\n",
       "      <td>San Antonio TX</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Milwaukee WI</th>\n",
       "      <td>Madison WI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Hamilton ON</th>\n",
       "      <td>Saint Catharines-Niagara ON</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Toledo OH</th>\n",
       "      <td>Windsor ON</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                    similar#1\n",
       "Jersey City NJ                      Newark NJ\n",
       "Columbus OH                     Cincinnati OH\n",
       "Victoria BC                        Seattle WA\n",
       "Buffalo NY        Saint Catharines-Niagara ON\n",
       "Calgary AB                         Seattle WA\n",
       "Norfolk VA                  Virginia Beach VA\n",
       "Mesa AZ                           Chandler AZ\n",
       "Fort Worth TX                    Arlington TX\n",
       "Arlington TX                    Fort Worth TX\n",
       "San Francisco CA                   Oakland CA\n",
       "Atlanta GA                      Birmingham AL\n",
       "Scottsdale AZ                         Mesa AZ\n",
       "Minneapolis MN                  Saint Paul MN\n",
       "Oklahoma City OK                     Tulsa OK\n",
       "Chicago IL                        New York NY\n",
       "Lexington KY                    Louisville KY\n",
       "Austin TX                      San Antonio TX\n",
       "Milwaukee WI                       Madison WI\n",
       "Hamilton ON       Saint Catharines-Niagara ON\n",
       "Toledo OH                          Windsor ON"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check the result\n",
    "\n",
    "most_similar_cities.sample(20)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1573f388",
   "metadata": {},
   "source": [
    "### Part 4: cluster similar cities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "c729a7c3",
   "metadata": {},
   "outputs": [],
   "source": [
    "data[\"num_searched\"] = data.cities.map(len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "2dec8b4c",
   "metadata": {},
   "outputs": [],
   "source": [
    "def search_distance(cities,similar2dist):\n",
    "    \"\"\"\n",
    "    input: \n",
    "        cities: a list of cities included in the search\n",
    "        similar2dist: a function which maps similarity to distance\n",
    "    return: \n",
    "        distance: distance among cities in the search\n",
    "    \"\"\"\n",
    "    sumdist = 0\n",
    "    total = len(cities)\n",
    "\n",
    "    # if total=1, then distance =0\n",
    "    for i1 in range(total-1):\n",
    "        city1 = cities[i1]\n",
    "\n",
    "        for i2 in range(i1+1,total):\n",
    "            city2 = cities[i2]\n",
    "\n",
    "            similarity = city_similarity.loc[city1,city2]\n",
    "            dist = similar2dist(similarity)\n",
    "\n",
    "            sumdist += dist\n",
    "\n",
    "    # if there are n cities in the search, sum will be calculated 0.5*n*(n-1) times\n",
    "    # then the distance biased too much to search which contains many cities\n",
    "    # so I decide to divide 'n', then the distance is O(n) instead of O(n2)\n",
    "    # which is more fair, but also takes account of \"more city in the search, more likely to be irrelevant\"\n",
    "    return sumdist / total"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "34e13466",
   "metadata": {},
   "outputs": [],
   "source": [
    "distances = data.cities.map(lambda cities: search_distance(cities,lambda s: np.sqrt(1-s*s)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "a9e71891",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.xticks(np.arange(0,5,0.5))\n",
    "# filter out distance=0, because a lot of search with one city will have distance=0\n",
    "\n",
    "distances[distances>0].hist(bins=100, stacked=True, figsize=(12,6))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "020208ce",
   "metadata": {},
   "source": [
    "I decide to choose distance=0.9 as the cutoff between 'high intent search' and 'low intent search'. if the distance is higher than 0.9, then I classify such search as 'low intent'."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "63c26725",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "percentage of low intent search: 12.21%\n"
     ]
    }
   ],
   "source": [
    "dist_cutoff = 0.9\n",
    "print('percentage of low intent search: {:.2f}%'.format((distances>dist_cutoff).mean() * 100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "4d7077f6",
   "metadata": {},
   "outputs": [],
   "source": [
    "low_intent_search = data.loc[distances>dist_cutoff,['num_searched','cities']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "85b4ecc5",
   "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>num_searched</th>\n",
       "      <th>cities</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>session_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>5AZ2X2A9BHH5U</th>\n",
       "      <td>3</td>\n",
       "      <td>[New York NY, Jersey City NJ, Philadelphia PA]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7BQ7VDHTZYRPD</th>\n",
       "      <td>3</td>\n",
       "      <td>[Vancouver BC, Victoria BC, Seattle WA]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>IGWYW7VQBR7BW</th>\n",
       "      <td>3</td>\n",
       "      <td>[Phoenix AZ, Chandler AZ, Mesa AZ]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>EEBNPTA77L6YT</th>\n",
       "      <td>4</td>\n",
       "      <td>[San Antonio TX, Corpus Christi TX, Arlington ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5HILT84J9UF4W</th>\n",
       "      <td>4</td>\n",
       "      <td>[Calgary AB, Seattle WA, Victoria BC, Portland...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>KU2EVVKG0EK9L</th>\n",
       "      <td>3</td>\n",
       "      <td>[San Diego CA, New York NY, Houston TX]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>H0ZSJHI32ERG9</th>\n",
       "      <td>3</td>\n",
       "      <td>[Houston TX, Chicago IL, Los Angeles CA]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CA656UDFEP1KP</th>\n",
       "      <td>3</td>\n",
       "      <td>[Los Angeles CA, Anaheim CA, Riverside CA]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47E5OO6JU2IXQ</th>\n",
       "      <td>4</td>\n",
       "      <td>[Montreal QC, Boston MA, Jersey City NJ, Newar...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1Y6XYIDGRIJMD</th>\n",
       "      <td>3</td>\n",
       "      <td>[Chicago IL, Milwaukee WI, Madison WI]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2445 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "               num_searched                                             cities\n",
       "session_id                                                                    \n",
       "5AZ2X2A9BHH5U             3     [New York NY, Jersey City NJ, Philadelphia PA]\n",
       "7BQ7VDHTZYRPD             3            [Vancouver BC, Victoria BC, Seattle WA]\n",
       "IGWYW7VQBR7BW             3                 [Phoenix AZ, Chandler AZ, Mesa AZ]\n",
       "EEBNPTA77L6YT             4  [San Antonio TX, Corpus Christi TX, Arlington ...\n",
       "5HILT84J9UF4W             4  [Calgary AB, Seattle WA, Victoria BC, Portland...\n",
       "...                     ...                                                ...\n",
       "KU2EVVKG0EK9L             3            [San Diego CA, New York NY, Houston TX]\n",
       "H0ZSJHI32ERG9             3           [Houston TX, Chicago IL, Los Angeles CA]\n",
       "CA656UDFEP1KP             3         [Los Angeles CA, Anaheim CA, Riverside CA]\n",
       "47E5OO6JU2IXQ             4  [Montreal QC, Boston MA, Jersey City NJ, Newar...\n",
       "1Y6XYIDGRIJMD             3             [Chicago IL, Milwaukee WI, Madison WI]\n",
       "\n",
       "[2445 rows x 2 columns]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "low_intent_search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "68ac304b",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}