SungryongHong-Klaviyo-Code-Screen.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Klaviyo Code Screen \n",
    "\n",
    "> This is **Sungryong Hong** (shongscience@gmail.com), an applicant for the data scientist position at **Klaviyo**. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Import basic libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# My Basic Libraries \n",
    "import sys\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "#from scipy.spatial import cKDTree\n",
    "import gc\n",
    "\n",
    "\n",
    "pd.set_option('display.max_rows', 500)\n",
    "pd.options.mode.chained_assignment = None\n",
    "#pd.set_option(\"display.precision\", 10)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Read and Explore the data using `pandas`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read data\n",
    "rawdf = pd.read_csv(\"./screening_exercise_orders_v201810.csv\", low_memory=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 13471 entries, 0 to 13470\n",
      "Data columns (total 5 columns):\n",
      "customer_id         13471 non-null int64\n",
      "gender              13471 non-null int64\n",
      "date                13471 non-null object\n",
      "value               13471 non-null float64\n",
      "predicted_gender    13471 non-null int64\n",
      "dtypes: float64(1), int64(3), object(1)\n",
      "memory usage: 526.3+ KB\n"
     ]
    }
   ],
   "source": [
    "rawdf.info()"
   ]
  },
  {
   "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>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>customer_id</th>\n",
       "      <td>1000</td>\n",
       "      <td>1001</td>\n",
       "      <td>1002</td>\n",
       "      <td>1003</td>\n",
       "      <td>1004</td>\n",
       "      <td>1005</td>\n",
       "      <td>1006</td>\n",
       "      <td>1007</td>\n",
       "      <td>1008</td>\n",
       "      <td>1009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gender</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <td>2017-01-01 00:11:31</td>\n",
       "      <td>2017-01-01 00:29:56</td>\n",
       "      <td>2017-01-01 01:30:31</td>\n",
       "      <td>2017-01-01 01:34:22</td>\n",
       "      <td>2017-01-01 03:11:54</td>\n",
       "      <td>2017-01-01 10:08:05</td>\n",
       "      <td>2017-01-01 15:42:57</td>\n",
       "      <td>2017-01-01 15:59:50</td>\n",
       "      <td>2017-01-01 18:01:04</td>\n",
       "      <td>2017-01-01 19:27:17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>value</th>\n",
       "      <td>198.5</td>\n",
       "      <td>338</td>\n",
       "      <td>733</td>\n",
       "      <td>772</td>\n",
       "      <td>508</td>\n",
       "      <td>338</td>\n",
       "      <td>1526.43</td>\n",
       "      <td>168</td>\n",
       "      <td>207.99</td>\n",
       "      <td>283</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>predicted_gender</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                    0                    1  \\\n",
       "customer_id                      1000                 1001   \n",
       "gender                              0                    0   \n",
       "date              2017-01-01 00:11:31  2017-01-01 00:29:56   \n",
       "value                           198.5                  338   \n",
       "predicted_gender                    1                    1   \n",
       "\n",
       "                                    2                    3  \\\n",
       "customer_id                      1002                 1003   \n",
       "gender                              1                    1   \n",
       "date              2017-01-01 01:30:31  2017-01-01 01:34:22   \n",
       "value                             733                  772   \n",
       "predicted_gender                    1                    1   \n",
       "\n",
       "                                    4                    5  \\\n",
       "customer_id                      1004                 1005   \n",
       "gender                              0                    1   \n",
       "date              2017-01-01 03:11:54  2017-01-01 10:08:05   \n",
       "value                             508                  338   \n",
       "predicted_gender                    1                    0   \n",
       "\n",
       "                                    6                    7  \\\n",
       "customer_id                      1006                 1007   \n",
       "gender                              1                    0   \n",
       "date              2017-01-01 15:42:57  2017-01-01 15:59:50   \n",
       "value                         1526.43                  168   \n",
       "predicted_gender                    1                    0   \n",
       "\n",
       "                                    8                    9  \n",
       "customer_id                      1008                 1009  \n",
       "gender                              0                    1  \n",
       "date              2017-01-01 18:01:04  2017-01-01 19:27:17  \n",
       "value                          207.99                  283  \n",
       "predicted_gender                    0                    1  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rawdf.head(10).transpose()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.1 Convert  the `date` column to `order_time`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 3.05 ms, sys: 377 µs, total: 3.42 ms\n",
      "Wall time: 3.14 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "rawdf['order_time'] = pd.to_datetime(rawdf.date)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>customer_id</th>\n",
       "      <td>1000</td>\n",
       "      <td>1001</td>\n",
       "      <td>1002</td>\n",
       "      <td>1003</td>\n",
       "      <td>1004</td>\n",
       "      <td>1005</td>\n",
       "      <td>1006</td>\n",
       "      <td>1007</td>\n",
       "      <td>1008</td>\n",
       "      <td>1009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gender</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <td>2017-01-01 00:11:31</td>\n",
       "      <td>2017-01-01 00:29:56</td>\n",
       "      <td>2017-01-01 01:30:31</td>\n",
       "      <td>2017-01-01 01:34:22</td>\n",
       "      <td>2017-01-01 03:11:54</td>\n",
       "      <td>2017-01-01 10:08:05</td>\n",
       "      <td>2017-01-01 15:42:57</td>\n",
       "      <td>2017-01-01 15:59:50</td>\n",
       "      <td>2017-01-01 18:01:04</td>\n",
       "      <td>2017-01-01 19:27:17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>value</th>\n",
       "      <td>198.5</td>\n",
       "      <td>338</td>\n",
       "      <td>733</td>\n",
       "      <td>772</td>\n",
       "      <td>508</td>\n",
       "      <td>338</td>\n",
       "      <td>1526.43</td>\n",
       "      <td>168</td>\n",
       "      <td>207.99</td>\n",
       "      <td>283</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>predicted_gender</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>order_time</th>\n",
       "      <td>2017-01-01 00:11:31</td>\n",
       "      <td>2017-01-01 00:29:56</td>\n",
       "      <td>2017-01-01 01:30:31</td>\n",
       "      <td>2017-01-01 01:34:22</td>\n",
       "      <td>2017-01-01 03:11:54</td>\n",
       "      <td>2017-01-01 10:08:05</td>\n",
       "      <td>2017-01-01 15:42:57</td>\n",
       "      <td>2017-01-01 15:59:50</td>\n",
       "      <td>2017-01-01 18:01:04</td>\n",
       "      <td>2017-01-01 19:27:17</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                    0                    1  \\\n",
       "customer_id                      1000                 1001   \n",
       "gender                              0                    0   \n",
       "date              2017-01-01 00:11:31  2017-01-01 00:29:56   \n",
       "value                           198.5                  338   \n",
       "predicted_gender                    1                    1   \n",
       "order_time        2017-01-01 00:11:31  2017-01-01 00:29:56   \n",
       "\n",
       "                                    2                    3  \\\n",
       "customer_id                      1002                 1003   \n",
       "gender                              1                    1   \n",
       "date              2017-01-01 01:30:31  2017-01-01 01:34:22   \n",
       "value                             733                  772   \n",
       "predicted_gender                    1                    1   \n",
       "order_time        2017-01-01 01:30:31  2017-01-01 01:34:22   \n",
       "\n",
       "                                    4                    5  \\\n",
       "customer_id                      1004                 1005   \n",
       "gender                              0                    1   \n",
       "date              2017-01-01 03:11:54  2017-01-01 10:08:05   \n",
       "value                             508                  338   \n",
       "predicted_gender                    1                    0   \n",
       "order_time        2017-01-01 03:11:54  2017-01-01 10:08:05   \n",
       "\n",
       "                                    6                    7  \\\n",
       "customer_id                      1006                 1007   \n",
       "gender                              1                    0   \n",
       "date              2017-01-01 15:42:57  2017-01-01 15:59:50   \n",
       "value                         1526.43                  168   \n",
       "predicted_gender                    1                    0   \n",
       "order_time        2017-01-01 15:42:57  2017-01-01 15:59:50   \n",
       "\n",
       "                                    8                    9  \n",
       "customer_id                      1008                 1009  \n",
       "gender                              0                    1  \n",
       "date              2017-01-01 18:01:04  2017-01-01 19:27:17  \n",
       "value                          207.99                  283  \n",
       "predicted_gender                    0                    1  \n",
       "order_time        2017-01-01 18:01:04  2017-01-01 19:27:17  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rawdf.head(10).transpose()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2 GroupBy `customer_id` "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8814"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(rawdf.customer_id.unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "grouped = rawdf.groupby('customer_id')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8814"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(grouped.groups.keys())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Questions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (A) Assemble a dataframe: \n",
    "\n",
    "> Assemble a dataframe with one row per customer and the following columns:\n",
    "    * customer_id\n",
    "    * gender\n",
    "    * most_recent_order_date\n",
    "    * order_count (number of orders placed by this customer)\n",
    "   Sort the dataframe by customer_id ascending and display the first 10 rows."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### A quick check for any inconsistent gender markers for each `customer_id`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    8814.0\n",
       "mean        1.0\n",
       "std         0.0\n",
       "min         1.0\n",
       "25%         1.0\n",
       "50%         1.0\n",
       "75%         1.0\n",
       "max         1.0\n",
       "Name: gender, dtype: float64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grouped.gender.nunique().describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> No error in gender labels; they are all unique and consistent. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Assemble `df` from `rawdf`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.concat([grouped.gender.agg(lambda x:x.value_counts().index[0]).rename('gender'),\\\n",
    "                grouped.order_time.agg('max').rename('most_recent_order_date'),\\\n",
    "                grouped.order_time.nunique().rename('order_count')],axis=1).reset_index()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> We do not need to find the most-frequent gender for each `customer_id` by `lambda x:x.value_counts().index[0]`, since we already know the labels are all consistent. But, if there are some errors in gender labels, this will be one of the most acceptable choices to fill each gender for `customer_id` group. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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>customer_id</th>\n",
       "      <th>gender</th>\n",
       "      <th>most_recent_order_date</th>\n",
       "      <th>order_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1000</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 00:11:31</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1001</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 00:29:56</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1002</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-02-19 21:35:31</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1003</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-04-26 02:37:20</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1004</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 03:11:54</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>1005</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-12-16 01:39:27</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>1006</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-05-09 15:27:20</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1007</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 15:59:50</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1008</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-12-17 05:47:48</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1009</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 19:27:17</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   customer_id  gender most_recent_order_date  order_count\n",
       "0         1000       0    2017-01-01 00:11:31            1\n",
       "1         1001       0    2017-01-01 00:29:56            1\n",
       "2         1002       1    2017-02-19 21:35:31            3\n",
       "3         1003       1    2017-04-26 02:37:20            4\n",
       "4         1004       0    2017-01-01 03:11:54            1\n",
       "5         1005       1    2017-12-16 01:39:27            2\n",
       "6         1006       1    2017-05-09 15:27:20            3\n",
       "7         1007       0    2017-01-01 15:59:50            1\n",
       "8         1008       0    2017-12-17 05:47:48            3\n",
       "9         1009       1    2017-01-01 19:27:17            1"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.sort_values(by=['customer_id']).head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (B) Plot the count of orders per week for the store"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "rawdf['order_week'] = rawdf.order_time.dt.week"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "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>customer_id</th>\n",
       "      <th>gender</th>\n",
       "      <th>date</th>\n",
       "      <th>value</th>\n",
       "      <th>predicted_gender</th>\n",
       "      <th>order_time</th>\n",
       "      <th>order_week</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1000</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 00:11:31</td>\n",
       "      <td>198.50</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 00:11:31</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1001</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 00:29:56</td>\n",
       "      <td>338.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 00:29:56</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1002</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 01:30:31</td>\n",
       "      <td>733.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 01:30:31</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1003</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 01:34:22</td>\n",
       "      <td>772.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 01:34:22</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1004</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 03:11:54</td>\n",
       "      <td>508.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 03:11:54</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>1005</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 10:08:05</td>\n",
       "      <td>338.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 10:08:05</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>1006</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 15:42:57</td>\n",
       "      <td>1526.43</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 15:42:57</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1007</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 15:59:50</td>\n",
       "      <td>168.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 15:59:50</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1008</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 18:01:04</td>\n",
       "      <td>207.99</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 18:01:04</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1009</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 19:27:17</td>\n",
       "      <td>283.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 19:27:17</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1010</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 19:41:40</td>\n",
       "      <td>1018.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 19:41:40</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>1011</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 22:04:54</td>\n",
       "      <td>596.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 22:04:54</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1012</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 22:08:16</td>\n",
       "      <td>151.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 22:08:16</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>1013</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 23:51:58</td>\n",
       "      <td>78.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 23:51:58</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>1014</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-02 00:28:42</td>\n",
       "      <td>98.50</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-02 00:28:42</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>1015</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:20:53</td>\n",
       "      <td>255.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:20:53</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>1016</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-02 01:24:13</td>\n",
       "      <td>311.39</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:24:13</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>1017</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:42:44</td>\n",
       "      <td>1000.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:42:44</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>1018</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:52:59</td>\n",
       "      <td>274.60</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:52:59</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>1017</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 02:08:52</td>\n",
       "      <td>186.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 02:08:52</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    customer_id  gender                 date    value  predicted_gender  \\\n",
       "0          1000       0  2017-01-01 00:11:31   198.50                 1   \n",
       "1          1001       0  2017-01-01 00:29:56   338.00                 1   \n",
       "2          1002       1  2017-01-01 01:30:31   733.00                 1   \n",
       "3          1003       1  2017-01-01 01:34:22   772.00                 1   \n",
       "4          1004       0  2017-01-01 03:11:54   508.00                 1   \n",
       "5          1005       1  2017-01-01 10:08:05   338.00                 0   \n",
       "6          1006       1  2017-01-01 15:42:57  1526.43                 1   \n",
       "7          1007       0  2017-01-01 15:59:50   168.00                 0   \n",
       "8          1008       0  2017-01-01 18:01:04   207.99                 0   \n",
       "9          1009       1  2017-01-01 19:27:17   283.00                 1   \n",
       "10         1010       1  2017-01-01 19:41:40  1018.00                 1   \n",
       "11         1011       1  2017-01-01 22:04:54   596.00                 1   \n",
       "12         1012       1  2017-01-01 22:08:16   151.00                 0   \n",
       "13         1013       0  2017-01-01 23:51:58    78.00                 0   \n",
       "14         1014       1  2017-01-02 00:28:42    98.50                 1   \n",
       "15         1015       0  2017-01-02 01:20:53   255.00                 0   \n",
       "16         1016       1  2017-01-02 01:24:13   311.39                 0   \n",
       "17         1017       0  2017-01-02 01:42:44  1000.00                 0   \n",
       "18         1018       0  2017-01-02 01:52:59   274.60                 0   \n",
       "19         1017       0  2017-01-02 02:08:52   186.00                 0   \n",
       "\n",
       "            order_time  order_week  \n",
       "0  2017-01-01 00:11:31          52  \n",
       "1  2017-01-01 00:29:56          52  \n",
       "2  2017-01-01 01:30:31          52  \n",
       "3  2017-01-01 01:34:22          52  \n",
       "4  2017-01-01 03:11:54          52  \n",
       "5  2017-01-01 10:08:05          52  \n",
       "6  2017-01-01 15:42:57          52  \n",
       "7  2017-01-01 15:59:50          52  \n",
       "8  2017-01-01 18:01:04          52  \n",
       "9  2017-01-01 19:27:17          52  \n",
       "10 2017-01-01 19:41:40          52  \n",
       "11 2017-01-01 22:04:54          52  \n",
       "12 2017-01-01 22:08:16          52  \n",
       "13 2017-01-01 23:51:58          52  \n",
       "14 2017-01-02 00:28:42           1  \n",
       "15 2017-01-02 01:20:53           1  \n",
       "16 2017-01-02 01:24:13           1  \n",
       "17 2017-01-02 01:42:44           1  \n",
       "18 2017-01-02 01:52:59           1  \n",
       "19 2017-01-02 02:08:52           1  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rawdf.head(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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>customer_id</th>\n",
       "      <th>gender</th>\n",
       "      <th>date</th>\n",
       "      <th>value</th>\n",
       "      <th>predicted_gender</th>\n",
       "      <th>order_time</th>\n",
       "      <th>order_week</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>13461</th>\n",
       "      <td>9286</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-12-31 20:58:40</td>\n",
       "      <td>158.60</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-12-31 20:58:40</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13462</th>\n",
       "      <td>9806</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-12-31 21:08:36</td>\n",
       "      <td>375.40</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-12-31 21:08:36</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13463</th>\n",
       "      <td>9807</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-12-31 21:24:08</td>\n",
       "      <td>237.35</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-12-31 21:24:08</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13464</th>\n",
       "      <td>9808</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-12-31 21:39:27</td>\n",
       "      <td>222.76</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-12-31 21:39:27</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13465</th>\n",
       "      <td>9809</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-12-31 21:39:33</td>\n",
       "      <td>606.40</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-12-31 21:39:33</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13466</th>\n",
       "      <td>9810</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-12-31 22:16:20</td>\n",
       "      <td>794.44</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-12-31 22:16:20</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13467</th>\n",
       "      <td>9811</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-12-31 22:24:06</td>\n",
       "      <td>148.60</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-12-31 22:24:06</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13468</th>\n",
       "      <td>9812</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-12-31 23:01:10</td>\n",
       "      <td>153.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-12-31 23:01:10</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13469</th>\n",
       "      <td>9813</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-12-31 23:11:02</td>\n",
       "      <td>534.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-12-31 23:11:02</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13470</th>\n",
       "      <td>8658</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-12-31 23:35:09</td>\n",
       "      <td>306.60</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-12-31 23:35:09</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       customer_id  gender                 date   value  predicted_gender  \\\n",
       "13461         9286       1  2017-12-31 20:58:40  158.60                 1   \n",
       "13462         9806       1  2017-12-31 21:08:36  375.40                 1   \n",
       "13463         9807       0  2017-12-31 21:24:08  237.35                 0   \n",
       "13464         9808       1  2017-12-31 21:39:27  222.76                 0   \n",
       "13465         9809       1  2017-12-31 21:39:33  606.40                 1   \n",
       "13466         9810       1  2017-12-31 22:16:20  794.44                 0   \n",
       "13467         9811       1  2017-12-31 22:24:06  148.60                 0   \n",
       "13468         9812       0  2017-12-31 23:01:10  153.00                 1   \n",
       "13469         9813       0  2017-12-31 23:11:02  534.00                 1   \n",
       "13470         8658       0  2017-12-31 23:35:09  306.60                 1   \n",
       "\n",
       "               order_time  order_week  \n",
       "13461 2017-12-31 20:58:40          52  \n",
       "13462 2017-12-31 21:08:36          52  \n",
       "13463 2017-12-31 21:24:08          52  \n",
       "13464 2017-12-31 21:39:27          52  \n",
       "13465 2017-12-31 21:39:33          52  \n",
       "13466 2017-12-31 22:16:20          52  \n",
       "13467 2017-12-31 22:24:06          52  \n",
       "13468 2017-12-31 23:01:10          52  \n",
       "13469 2017-12-31 23:11:02          52  \n",
       "13470 2017-12-31 23:35:09          52  "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rawdf.tail(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Due to the duplicated week labels of `52`, let's replace the week labels for all orders on 2017-01-01 with `0`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "rawdf['order_week'].loc[rawdf.order_time < '2017-01-02'] = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "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>customer_id</th>\n",
       "      <th>gender</th>\n",
       "      <th>date</th>\n",
       "      <th>value</th>\n",
       "      <th>predicted_gender</th>\n",
       "      <th>order_time</th>\n",
       "      <th>order_week</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1000</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 00:11:31</td>\n",
       "      <td>198.50</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 00:11:31</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1001</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 00:29:56</td>\n",
       "      <td>338.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 00:29:56</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1002</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 01:30:31</td>\n",
       "      <td>733.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 01:30:31</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1003</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 01:34:22</td>\n",
       "      <td>772.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 01:34:22</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1004</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 03:11:54</td>\n",
       "      <td>508.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 03:11:54</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>1005</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 10:08:05</td>\n",
       "      <td>338.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 10:08:05</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>1006</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 15:42:57</td>\n",
       "      <td>1526.43</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 15:42:57</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1007</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 15:59:50</td>\n",
       "      <td>168.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 15:59:50</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1008</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 18:01:04</td>\n",
       "      <td>207.99</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 18:01:04</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1009</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 19:27:17</td>\n",
       "      <td>283.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 19:27:17</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1010</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 19:41:40</td>\n",
       "      <td>1018.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 19:41:40</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>1011</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 22:04:54</td>\n",
       "      <td>596.00</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 22:04:54</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1012</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 22:08:16</td>\n",
       "      <td>151.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 22:08:16</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>1013</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 23:51:58</td>\n",
       "      <td>78.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 23:51:58</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>1014</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-02 00:28:42</td>\n",
       "      <td>98.50</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-02 00:28:42</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>1015</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:20:53</td>\n",
       "      <td>255.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:20:53</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>1016</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-02 01:24:13</td>\n",
       "      <td>311.39</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:24:13</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>1017</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:42:44</td>\n",
       "      <td>1000.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:42:44</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>1018</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:52:59</td>\n",
       "      <td>274.60</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 01:52:59</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>1017</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 02:08:52</td>\n",
       "      <td>186.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-02 02:08:52</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    customer_id  gender                 date    value  predicted_gender  \\\n",
       "0          1000       0  2017-01-01 00:11:31   198.50                 1   \n",
       "1          1001       0  2017-01-01 00:29:56   338.00                 1   \n",
       "2          1002       1  2017-01-01 01:30:31   733.00                 1   \n",
       "3          1003       1  2017-01-01 01:34:22   772.00                 1   \n",
       "4          1004       0  2017-01-01 03:11:54   508.00                 1   \n",
       "5          1005       1  2017-01-01 10:08:05   338.00                 0   \n",
       "6          1006       1  2017-01-01 15:42:57  1526.43                 1   \n",
       "7          1007       0  2017-01-01 15:59:50   168.00                 0   \n",
       "8          1008       0  2017-01-01 18:01:04   207.99                 0   \n",
       "9          1009       1  2017-01-01 19:27:17   283.00                 1   \n",
       "10         1010       1  2017-01-01 19:41:40  1018.00                 1   \n",
       "11         1011       1  2017-01-01 22:04:54   596.00                 1   \n",
       "12         1012       1  2017-01-01 22:08:16   151.00                 0   \n",
       "13         1013       0  2017-01-01 23:51:58    78.00                 0   \n",
       "14         1014       1  2017-01-02 00:28:42    98.50                 1   \n",
       "15         1015       0  2017-01-02 01:20:53   255.00                 0   \n",
       "16         1016       1  2017-01-02 01:24:13   311.39                 0   \n",
       "17         1017       0  2017-01-02 01:42:44  1000.00                 0   \n",
       "18         1018       0  2017-01-02 01:52:59   274.60                 0   \n",
       "19         1017       0  2017-01-02 02:08:52   186.00                 0   \n",
       "\n",
       "            order_time  order_week  \n",
       "0  2017-01-01 00:11:31           0  \n",
       "1  2017-01-01 00:29:56           0  \n",
       "2  2017-01-01 01:30:31           0  \n",
       "3  2017-01-01 01:34:22           0  \n",
       "4  2017-01-01 03:11:54           0  \n",
       "5  2017-01-01 10:08:05           0  \n",
       "6  2017-01-01 15:42:57           0  \n",
       "7  2017-01-01 15:59:50           0  \n",
       "8  2017-01-01 18:01:04           0  \n",
       "9  2017-01-01 19:27:17           0  \n",
       "10 2017-01-01 19:41:40           0  \n",
       "11 2017-01-01 22:04:54           0  \n",
       "12 2017-01-01 22:08:16           0  \n",
       "13 2017-01-01 23:51:58           0  \n",
       "14 2017-01-02 00:28:42           1  \n",
       "15 2017-01-02 01:20:53           1  \n",
       "16 2017-01-02 01:24:13           1  \n",
       "17 2017-01-02 01:42:44           1  \n",
       "18 2017-01-02 01:52:59           1  \n",
       "19 2017-01-02 02:08:52           1  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rawdf.head(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "weekly_orders = rawdf.order_week.value_counts().sort_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x374.4 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot settings\n",
    "plt.rc('font', family='serif') \n",
    "plt.rc('font', serif='Times New Roman') \n",
    "plt.rcParams.update({'font.size': 16})\n",
    "plt.rcParams['mathtext.fontset'] = 'stix'\n",
    "\n",
    "\n",
    "fig = plt.figure(figsize=(8,5.2))\n",
    "\n",
    "#plt.axis([0,lenmax,0,1.0])\n",
    "#plt.title(\"Number of Citations in Baltimore\")\n",
    "#plt.yscale('log')\n",
    "#plt.xscale('log')\n",
    "plt.ylabel(r'Number of Weekly Orders')\n",
    "plt.xlabel(r'Weeks in 2017')\n",
    "plt.plot(weekly_orders.index.values,weekly_orders.values,color='b')\n",
    "\n",
    "\n",
    "plt.tight_layout(pad=0.4, w_pad=0.5, h_pad=1.0)\n",
    "#fig.savefig(\"car-accidents.eps\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (C) Compute the mean order value for gender 0 and for gender 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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>customer_id</th>\n",
       "      <th>gender</th>\n",
       "      <th>most_recent_order_date</th>\n",
       "      <th>order_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1000</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 00:11:31</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1001</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 00:29:56</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1002</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-02-19 21:35:31</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1003</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-04-26 02:37:20</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1004</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 03:11:54</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   customer_id  gender most_recent_order_date  order_count\n",
       "0         1000       0    2017-01-01 00:11:31            1\n",
       "1         1001       0    2017-01-01 00:29:56            1\n",
       "2         1002       1    2017-02-19 21:35:31            3\n",
       "3         1003       1    2017-04-26 02:37:20            4\n",
       "4         1004       0    2017-01-01 03:11:54            1"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "groupedGender = df.groupby('gender')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Mean value for each gender group"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.5334694803721352"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "groupedGender.get_group(0).order_count.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.5228046289993193"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "groupedGender.get_group(1).order_count.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Statistical Significance \n",
    "\n",
    "> In astronomy, we generally use the Kolmogorov–Smirnov (K-S) test to determine whether two samples are statistically different from each other or not. Let's do a K-S test for these two gender groups. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy import stats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Ks_2sampResult(statistic=0.01837985023825739, pvalue=0.4427840079013693)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "stats.ks_2samp(groupedGender.get_group(0).order_count.values,groupedGender.get_group(1).order_count.values)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> **Due to the high `p` value, I do not think that the difference is significant.**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (D) Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "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>customer_id</th>\n",
       "      <th>gender</th>\n",
       "      <th>date</th>\n",
       "      <th>value</th>\n",
       "      <th>predicted_gender</th>\n",
       "      <th>order_time</th>\n",
       "      <th>order_week</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1000</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 00:11:31</td>\n",
       "      <td>198.5</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 00:11:31</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1001</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 00:29:56</td>\n",
       "      <td>338.0</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 00:29:56</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1002</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 01:30:31</td>\n",
       "      <td>733.0</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 01:30:31</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1003</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 01:34:22</td>\n",
       "      <td>772.0</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 01:34:22</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1004</td>\n",
       "      <td>0</td>\n",
       "      <td>2017-01-01 03:11:54</td>\n",
       "      <td>508.0</td>\n",
       "      <td>1</td>\n",
       "      <td>2017-01-01 03:11:54</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   customer_id  gender                 date  value  predicted_gender  \\\n",
       "0         1000       0  2017-01-01 00:11:31  198.5                 1   \n",
       "1         1001       0  2017-01-01 00:29:56  338.0                 1   \n",
       "2         1002       1  2017-01-01 01:30:31  733.0                 1   \n",
       "3         1003       1  2017-01-01 01:34:22  772.0                 1   \n",
       "4         1004       0  2017-01-01 03:11:54  508.0                 1   \n",
       "\n",
       "           order_time  order_week  \n",
       "0 2017-01-01 00:11:31           0  \n",
       "1 2017-01-01 00:29:56           0  \n",
       "2 2017-01-01 01:30:31           0  \n",
       "3 2017-01-01 01:34:22           0  \n",
       "4 2017-01-01 03:11:54           0  "
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rawdf.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "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>predicted_gender</th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gender</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3349</td>\n",
       "      <td>3410</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1463</td>\n",
       "      <td>5249</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "predicted_gender     0     1\n",
       "gender                      \n",
       "0                 3349  3410\n",
       "1                 1463  5249"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.crosstab(rawdf.gender,rawdf.predicted_gender)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "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>predicted_gender</th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gender</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.495487</td>\n",
       "      <td>0.504513</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.217968</td>\n",
       "      <td>0.782032</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "predicted_gender         0         1\n",
       "gender                              \n",
       "0                 0.495487  0.504513\n",
       "1                 0.217968  0.782032"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.crosstab(rawdf.gender,rawdf.predicted_gender).apply(lambda row: row/row.sum(), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "cmatrix = pd.crosstab(rawdf.gender,rawdf.predicted_gender).apply(lambda row: row/row.sum(), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "import seaborn as sbn\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZQAAAF7CAYAAAD11333AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzt3Xl8FdX9//FXgCQkLCJqQVBRqn4EFwQVd0WrVqutW9Uu+lNxq0sVWm1VqqLighXESkUUFa1b/brx/dW6/9zaiuKGCvpxAaQsKogohEBCcn9/nAleLpckN5lkkpv308c8knvmzMy5l+t88jnnzExBKpVCRESksdol3QAREckPCigiIhILBRQREYmFAoqIiMRCAUVERGKhgCIiIrHokHQDJDlm1gU4E/glsA3h+zADmARMcvfqJjx2AXA9cDrQEbjI3W+Ncf+TgZPdvSCufdbjmKcAd0cvj3X3x9ZT72bgfOBzd9+yAccpAjZx9/n1bM8B7v5SrscRyZUylDbKzAx4E7gOeB+4FLgcWAlMBO6NTvpN5XDgD8BrwAXACzHvfyJwUsz7zMWRDVxXKzPrQ/j3Orge1V8hfAYfNvR4IrlQhtIGmVlHYAqwMbCru7+XtnqMmf0VOAd4A/hLEzVjp+jnJe7+ftw7d/fXCMEqCbOBw82svbtXpa8ws0FAH2BRA/e9FbBtfSq6+yxgVgOPI5IzZSht0zmAAcMzgkmNC4FvgN80YRuKop/LmvAYSZkCbATsk2Xd0cBnwMxmbZFIM1CG0jb9AlgOPJhtpbuXm9nuwOfp5Wa2L3AFsEdU9AYw0t1fSaszB3ga+BdwCfBD4L/AOHf/a1qdPtEms83sc3ffMiqf4+5DMo67VrmZbQjcBBwI9ADmAQ8DV7r7yqjOZDLGUKLuolHAoUAXwIHx7n5HWp3J0fs7CbgR2I0Q9P4O/NHdy7N9ZhmeJYxN/Qx4OWPd0cBjwODMjczsQOCiaF1X4CvgH9Fxl2aM0dxtZne7e4GZjQQuJoyFTQA6AcOAqqj+AYTur1ejfe9S84eEmW0NTAfeAfZrynEzyX/KUNqYaFxkIPCWu1eur567f+LuFWnb/Qx4CdgCuDpatgBeiNalO4zQVfYIMBwoA8ab2U+i9cOAx6Pfh0evc/EwcARwB3Bu1K6LqaV7zsy2AqYRxi/uIJy4lwC3m9kNGdV/QAgKHxHGd/4N/Ba4sp7tK4+2X2usxMy2Abbn+/eevu4Q4DlCMLicMGj/BiEw3RRVewW4Nvr9dtYeIyokTKa4iRAI/5W+/yhQDAVWAxPMrMDM2gF3ASlC8FUwkUZRhtL2bEz4d19Y3w3MrAPwV2A+Yczlu6h8IvABcKuZPZUWoDYHdk77K/hxYAHwa+Cf7v6Eme1M+Gv9CXefk0NbfgAcRJgVdmNUPCkKlH1r2fQ6QjfUbu7+drSvvxK6py40s3vcfUZUd0PgfHe/JXp9h5nNjNr/h3o29QngKDPbPm2/RxM+96lZ6g8nZHIHpQXyCWb2GnAscKq7zzKz5wgTKF5z9/vStm8H3OLuo2sKzGzP9AO4u0fZzPXAqUBnYF/gN+7+WT3fl8h6KUNpe2oGidvnsM0gYDNC99B3NYXuvhQYD/QGdk2r7+ljM+7+BfAl0LOhjU7zLaG77hwzO9bMOkXHGOruB2XbwMzaE2aVPVMTTKJtqoFrgAJC91S6hzNeTyd0r9XXPwifdXqWUhNAs93i+whCV1R6VrgR8B3hxF8fz9Sjzo2E2X3XE9770+4+sZ77F6mVAkrb8w1QQejWqa+top+eZV3NlNQ+aWXZZjCtIrcglpW7rwLOIpzcHwG+NrNnzOzMaPZaNhsTTsr1bT+s+x5yar+7f03odvoZgJltCuxOGD/JVr8K6Gtmt5rZy2Y2D1gMHEIIePXxVT3aVQWcBmxCuP7n9HruW6ROCihtTPTX8WvALlFXVlZmNsrMHjSzntR+Qqv5DlWklcXdF7/WidzdHyB0q50GPEkYRJ8ITDWz4izb59r+muylsZ4ABkfB5ChgKWG8Zx1mdhZhzORA4GPgBsL7uj+H41XVXQX4fvZZB9bNzEQaTAGlbXqM8Bf7CdlWmlkJ4S/Xg4CvgTnRqu2yVY9+/jeGdlUBawWEKOhtnPa6s5ntA6Tc/S53P5bw1/bNwADCX/SZFhEmBjR1+zM9QQhmRxACyv9199XrNCBkVmOBF4Ed3P0Md/+Lu79ObplkncxsC0J319OELrIbotlvIo2mgNI23U6YEjzGzHZIXxGNN0wgdCmNjgba3yIMJp9jZl3T6nYlXNOyMKrTWF+E3VpJWtnPCF0zNXYgTH89raYgGnd4J3q5zl/pUTfPU8Ah0YWFNe0vAP5ImOX0ZAztzzzuHMLYyy+BIaynuwsoAUqBj9MDTjRxYf/o95pssub9NfT/3dsJmcm5wHmE2WG3N3BfImvRLK82yN1XmtnRhKmt08zsfsKU2o2A44Cdgf8h/NWMu1ea2W8JA9VvmtmkaFenA72An8fURfQgcAvwtJndB2xNmDabfj3M64SAck301/Z7hO6v3xKm+T6/nn1fTOhOesnMbiEEwaOjsrHu3lQXGj5BuHanjPB5r8PdvzGz14GhZvYdYaxnB8LnW/O5diGMf9WM7ZwYBcR76tsQMzsV+DFwaXQVPWZ2HTDSzE5z9ztzfXMi6ZShtFHu/g4hcIwH9iTM/hlBuJfXUOCE9CDh7o8SupMWEE6QlxJuMXKAuz8RU7Nujfa9FSGwDCGc9D9Ia0eK0H10G6EraTwh6DwataWCLKJpsbsD/yTcAeAGoBtwmrv/Pqb2Z1Pz2Txdx0WRxxGmMA8FxhHu1XU9YaoyhMCHu39E+Gx2jerVq7sqGscZS7hC/8a0VdcDnxCy1d712ZfI+hSkUtlmMIqIiORGGYqIiMRCAUVERGKhgCIiIrFQQBERkVgooIiISCxa9XUoU3sdoylq0mDFHep7pxKR7AbOnRLLY7IrF89q8LmscOO+Tfmo7pwoQxERkVi06gxFRCQvVOdHtqyAIiKStFR+PCxTAUVEJGnVCigiIhKDlDIUERGJRZ5kKJrlJSIisVCGIiKSNHV5iYhILDRtWEREYqEMRUREYpEng/IKKCIiCcuXacOa5SUiIrFQhiIikjR1eYmISCzypMtLAUVEJGmaNiwiIrFQhiIiIrHQGIqIiMQiTzIUTRsWEZFYKEMREUmaurxERCQOqZRmeYmISBzyZAxFAUVEJGnq8hIRkVjkSYaiWV4iIhILZSgiIknTrVdERCQWedLlpYAiIpI0DcqLiEgslKGIiEgs8iRD0SwvERGJhTIUEZGk5UmGooAiIpIw3ctLRETioQxFRERioVleIiISizzJUDTLS0REYqEMRUQkaeryEhGRWDRTl5eZDQBuA3YCZgFD3X1aRp1fAxMzNi0BXnD3Q2rbvwKKiEjSmiFDMbMiYAowDtgPOBZ41sz6uPt3NfXc/X7g/rTtBgLPAhfVdQwFFBGRpDVPhjIEKHT3cdHrh8zsPOAE4I5sG5hZISG4jHT36XUdQAFFRCRpjQgoZtYN6JZl1VJ3X5r2uj/wYUadj4Ada9n9uUA5cGt92qJZXiIirdswYHaWZVhGvc7AioyyFUBptp1GXWQXEbKTVH0aogxFRCRpjRtDGQdMzlK+NON1GWFwPV0psHw9+z0UqAaerG9DFFBERJLWiC6vqFsrM3hkMxMYnlG2HXDveuofCTzs7vVunAKKiEjSmuc6lBeBAjMbDownzPLaCXh8PfX3AC7L5QAaQxERSVp1dcOXenL3CuAwQiBZAowAjnL3RWZ2qZnNyNhkS2BBLm9DGYqISNKa6Up5d/8A2CdL+bXAtRllnXLdvzIUERGJhTIUEZGk5cndhhVQRESSpoAiIiKxSNXrusEWTwFFRCRpylBERCQWCigiIhKLPHnAlqYNi4hILJShiIgkTV1eIiISC83yEhGRWChDERGRWCigiIhILDTLS0RE5HvKUEREEpaq1qC8iIjEQWMoIiISizwZQ1FAERFJmrq8REQkFnnS5aVZXiIiEgtlKK1Z+3b0ueIUNj56Xwrat2fRoy8z96p7SFWuXqdqYY8N2eWdO9cpf3OHk1m9ZFlztFZamvbt6H3ZUDY8cj8KOrRnyWMvsWDU3Vm/Px16dGfHaXevU/7egBOp+kbfn0bLkwxFAaUV2+KSX9PtgEH4KddRUNiBrW8+n1TlauZedc86dUu23ZzVS5czff/z1ypXMGm7ev3xJLoOGcSs066hXWEHthh7AamKShZcM3mduiXbbM7qb5fz4YHnrlWuYBKTPLmXl7q8WqmC4kJ6nHwon181meVvfcyyqTOZ/adJ9DjpEAo6Fq1Tv3TbzSn/dD6Vi5autUjbVFBcyMYnHcb8UXez4m1n+eszmHfFHWx84qEUFK/7/em47eas+nQeqxctXWuRmFRXN3xpQZotQzGzzsAJwI5AKbAcmAk87u5fN1c78kWn7beifacSlk2duaZs2WszaN+phE7bb8nytz5eq36JbU75Z/Obu5nSQpX0D9+f5a/PWFO2fOoHtO9UQsn2W7HibV+rfsdtt2Clvj9NJ09meTVLhmJmuwJzgVOATkA50AU4GfjMzPZvjnbkk6Ke3akqK6dq2Yo1ZVXLy6lasZKiTTdap36pbUFRz+7s8I/rGfTOndg9l1C8Zc/mbLK0IIU9N6KqrJzqtO9PdS3fn47bbkFhz+5sO+XP7PDm3fS9awRFffT9iU2quuFLC9JcGcqdwNnu/vfMFWb2C+AvwIBmakteaFdSTPWqdQdPU6sqaVdUuE55yda9Kf9sAZ9fNZnU6mp6nX8s2z96NdMPuICq71asU1/yW7uSYlIVleuUpyoqKcjy/em49Was+mw+80fdRWp1FT3PO45t/udaPjrot1R9V9YcTZZWoLkCyg+BR9az7hHg9mZqR96oXllBu+J1//kKigupKl+1Tvnbu55BqqqaVEUIQp/8ZgyDpk2k+xF7seiB55u8vdKypFZWZA0cBUWFVGf5/nwweChUf//9mX3On9l+6iS6/WQvvn7ouSZvb95Tl1dO3gPOWc+6c4HpzdSOvFGx8GvadyqhXaeOa8rady6hfWlHKr9Ysk796vKKNScDCCeUVXO/ojhL94bkv4qFi6PvT8masna1fH9SKzO+P6sqqPjvlxTq+xOLVHV1g5eWpLkCytnA78zsv2b2rJk9YWbPmNkc4HfAWc3UjrxRNnMOVWXldN29/5qyLntuT1VZOWUz56xVt7Bnd3b96G90GdxvTVm7Th3p2LcX5Z/Ma64mSwtSHn1/Og/+/vvTeY8dqCorp/zD2WvVLezRnZ0+eIBOaXXbdSqheKterPxU359YVKcavrQgzRJQ3H06sC0wFHgSmAY8BZwJbO3uM2vZXLJIrazgq/ufZ8trTqfL4H502b0/W406nS8nP01qVSXtu5TSoXtXACq/WELZ+7PYctRpdB64DaX9+rDNrb+jctE3LPnn1ITfiSQhtaqCrx98js2uPpNOg/vTaXB/NrvyDBbf+88wDpf+/flyCSs+mMVmI8+gdOdt6bhdH7Yc/3tWL1rK0qdeS/id5Ik8GZQvSLXiC2qm9jqm9TY+BgVFHdjyqtPY6Mh9SFVVsfixV/j8yslQVc0PbzqPrnvtwDu7/waADt270ufyk+l24CDalRbz7avvMedPd1Ixf1GybyJBxR2qkm5CogqKOrDZyDPo9tN9oKqaJU+8zPyr74KqarYYcz6d99iBmXufCYTvT68/nUrXIYNoV9qRZf+azrwr7qCyDX9/AAbOnVIQx37Krvp1g89lnS6/P5Y2xEEBRdqsth5QpPEUUNamW6+IiCSthQ2uN5QCiohI0lrY4HpDKaCIiCSthQ2uN5QCiohI0pShiIhIHFraBYoNpYAiIpK0PMlQ9DwUERGJhTIUEZGk5UmGooAiIpI0zfISEZFYKEMREZE4pBRQREQkFgooIiLSmpjZAOA2YCdgFjDU3adlqbcpMAE4AFgJ3O7ul9W1f00bFhFJWnV1w5d6MrMiYArwd6AbcA3wrJl1zVJ9CrAQ6AHsAZxsZr+q6xjKUEREktY8XV5DgEJ3Hxe9fsjMzgNOAO6oqWRmuwN9gb3dvRKYbWZDgPK6DqCAIiKStEYEFDPrRsg4Mi1196Vpr/sDH2bU+QjYMaNsF+B9YKSZnULo8rrV3cfU1RZ1eYmIJCyVSjV4AYYBs7MswzIO0xlYkVG2AijNKOsO7AtUEjKVY4AL1eUlItIaNK7LaxwwOUv50ozXZUBJRlkpsDyjbBXwnbuPjF5PN7NJhMDyQG0NUUAREWnFom6tzOCRzUxgeEbZdsC9GWUfAaVmVuTuFVFZvWKFAoqISNKaZ1D+RaDAzIYD44FjCdOHH8+o9xywCBhjZr8HDDgNOLuuA2gMRUQkYanqVIOX+oqyjcMIgWQJMAI4yt0XmdmlZjYjqrcS2J8wfrIQeBq4wd0fresYBdGgTqs0tdcxrbfxkrjiDlVJN0FauYFzpxTEsZ9vT/5Rg89lG9zzQixtiIO6vEREkpYfNxtWQBERSVq+3BxSYygiIhILZSgiIknLkwxFAUVEJGkaQxERkTjkyxiKAoqISNKUoYiISBzyJUPRLC8REYmFMhQRkaSpy0tEROKQUkAREZFYKKCIiEgclKGIiEg88iSgaJaXiIjEQhmKiEjC1OUlIiKxUEAREZFYKKCIiEg8Ui3mKb6NooAiIpIwZSgiIhKLVHV+ZCiaNiwiIrFQhiIikjB1eYmISCxSGpQXEZE45EuGUu8xFDPbuSkbIiLSVqWqCxq8tCS5DMo/Y2a7NVlLRETaqFSq4UtLkktA+QYobqqGiIhI65bLGMr/BZ4ys/8FZgHl6Svd/do4GyYi0la0tK6rhsoloPwcWAzsFS3pUoACiohIA7S5gOLuWzVlQ0RE2qqWNhbSUDlPGzaz3YF+wKPA5sDH7r467oaJiLQVbS5DMbMNgMeAAwhdXK8Ao4Gtzexgd5/XNE0UEclv+XJhYy6zvEYTAtDmwIqo7HzgW+DGmNslIiKtTC4B5XDgInefX1Pg7rOB84Afxd0wEZG2IlXd8KUlyWUMpTuwJEv5SqAknuaIiLQ91W2wy+tV4PS01ykzaw9cDLwWa6tERNqQVKqgwUtLkkuGciHwkpkdQLhi/hbCbK8NgYOaoG0iIm1CvszyqneG4u4fAAOA54BngQrgAaCfu7/dNM0TEcl/+XIvr5yuQ4kG5P/URG0REZFWrNaAYmZ31XdH7j608c0REWl78qXLq64MZfO039sDQ4D5wNuELq+BQB/CBY8iItIA+TLLq9aA4u4H1/xuZmOAz4Ez3b0yKisAxgOdmrKRIiL5rKXN1mqoXKYNnw5cXxNMANw9BdxMuBOxiIg0QFsclC8D+gOeUT4Y+Dq2FomItDHN1eVlZgOA24CdCM+1Guru07LUGwpMBFalFZ/r7vfUtv9cAsodwJ1m1o8whlJAeC7KBcDlOexHRESamZkVAVOAccB+wLHAs2bWx92/y6g+CBjj7hfncoxcAspIYDXwW6BHVDYfuMzdb87loCIi8r1mGkMZAhS6+7jo9UNmdh5wAiFhSLcLYTgjJ7k8YCsFXA1cbWYbAyl3V1eXiEgjNdNYSH/gw4yyj4Ad0wuiW2rtBJxkZmMJd5efBIyO4sB65XRho5ltThgzKQIKzGzNOnd/IJd9xWGfxa839yElj5QveDXpJogAjRtDMbNuQLcsq5a6+9K01535/tEjNVYApRllmwBvAvcAxxBusTUF+A64tba25PKArTOinbXPsjpFuA2LiIjkqJFdXsOAK7KUX0kYqqhRxrp3hi8FlqcXuPsXwP5pRe+a2S2EMZd4AgrwO+BuwjNRvs1hOxERqUUjZ3mNAyZnKV+a8XomMDyjbDvg3vQCM9seON7d04NUEeFRJbXKJaD0AY5UMBERiVdjhlCibq3M4JHNi4ShiuGEC9KPJYyVPJ5RbynwezObB9xJuCPK+YSHKdYqlwsbp5ExeCMiIq2Du1cAhxECyRJgBHCUuy8ys0vNbEZUbz7wM+AswrjJo8DV7v5IXccoSNVzeoGZnQpcT4hYn7D2BS+JDMp3KOrdwq4TldZEg/LSWIUb941lvu9/Nj22weeyvRY+2mLu25JLl9ed0c9sF7poUF5EpIHy5V5euVyHkkv3mIiI1FN10g2ISU7XoQCYWQ/CvOTXgS7u/lXsrRIRaUNStLEMxcyKgQnAKYSAui0wxsy6Asdo9peISMNU58locC7dWFcAuwL78P185BsID+G6IeZ2iYhIK5NLQDkeON/d/0M0bdrdXwPOIEwxExGRBqimoMFLS5LLGEovwhMbM30BbBBPc0RE2p58GUPJJUN5Bzgq7XVNr9+ZwPTYWiQi0sZUN2JpSXLJUC4GnjazPQj3dbk4etjWHsBPmqJxIiJtQZvLUNz9VWBvoIJwpfxuwBxgkLs/3yStExFpA/IlQ6lXQDGzwwDc/V3CrY7fIHRzrSYEFhERaeNqDShmVmxmzwFPmFnfqPhEYBtgM8JzhyeY2Q+btpkiIvmrrWQovwP6Av3cfVZa+cnufjDh6Y1OeM68iIg0QIqCBi8tSV0B5ZfAiIxgskZ0O+TRwKFxN0xEpK2oLmj40pLUNctra+A/GWWfEQbma7wGbBFno0RE2pKWdoFiQ9UVUFYCHdML3H3njDolZDyTWERE6i9PbuVVZ5fXR8CP6qjzY+C9eJojIiKtVV0B5V7gCjPrn22lmW1HeIzk3XE3TESkrciXWV51dXlNJNxu5W0zm0x4yP1ioDuwL3Aq8Ly739+UjRQRyWfVBW1gDMXdU2Z2OHARcDbhvl01viQ8Y/76pmueiEj+y5cxlDrv5eXuVUSBw8y2AnoAXwOfuXtLy7hERFqdfDmR5vQIYHefDcxuoraIiLRJLe16kobK5fb1IiIi65VThiIiIvFrKxc2iohIE2szg/IiItK08mUMRQFFRCRhbXKWl4iIxC9furw0y0tERGKhDEVEJGEaQxERkVhoDEVERGKhgCIiIrFIqctLRETioAxFRERikS8BRdOGRUQkFspQREQSli8XNiqgiIgkTNehiIhILPJlDEUBRUQkYQooIiISi3wZQ9EsLxERiYUyFBGRhGlQXkREYqExFBERiUW+jKEooIiIJKy6mUKKmQ0AbgN2AmYBQ919Wi31uwHvAZe7++S69q9BeRGRhFU3YqkvMysCpgB/B7oB1wDPmlnXWja7Dehd32MooIiItA1DgEJ3H+fule7+EDADOCFbZTM7GegKvF/fA6jLS0QkYY3p8Iq6pbplWbXU3Zemve4PfJhR5yNgxyz73Aq4AtgLeLq+bVGGIiKSsEZ2eQ0DZmdZhmUcpjOwIqNsBVCaXmBm7YH7gAvd/Ytc3ocyFBGRhDXyOpRxwOQs5UszXpcBJRllpcDyjLLLAHf3x3JtiAKKiEjCGjPLK+rWygwe2cwEhmeUbQfcm1H2C6CXmR0Tve4C3Gpmg939nNoOoIAiIpKwZroO5UWgwMyGA+OBYwnThx9Pr+Tu26W/NrN3gXGaNiwiIgC4ewVwGCGQLAFGAEe5+yIzu9TMZjT2GAWpVOu9RrNDUe/W23hJXPmCV5NugrRyhRv3jeUuXJds+asGn8uum/NAi7kTmLq8REQS1lxXyjc1BRQRkYTlRzhRQBERSZzuNiwiIrHIly4vzfISEZFYKEMREUlYfuQnCigiIonTGIqIiMQilSc5igKKiEjC8iVD0aB8K9K+fXvGjrmShfPfY9GXM7hp7FUUFhZmrdut2wbcNuEG/vv523y58AP+/tDt9OrVc8363r035aEHJ/LFgveZN/cdbp94IxtsUNuD26S1W726iuvG3ca+h5/AXocex7U3TaCysnKdem+8/R477H1Y1mXKU88DsLysjMuuu4l9Dz+BfQ8/gRGjxvDtd8ua+y3ljWpSDV5aEmUorcg1oy7m0EMP5OhjTqWoqJC77hxHZUUlf7j46nXq3jlpLL169eSYY4dSvnIlN1x/GY8/djd77nU4AI89eheLvlrMQYccR0nHjowffx133zWOY44d2txvS5rJzRMn86+pb3LL6JFUVlYyYtQYCgs7cNF5Z6xVb+CO/Xjpf+9fq2z8pL8x9c13+dF+ewJwzdgJzJozl9vGjAJg5OibueL6mxl37Z+a583kmZYVFhpO9/JqJYqLi/ly4fuceNK5/OPJ5wA44vCDue9vf6Vnr51YuXLlmrobbbQhXy78gL32PoI3pr0DQK9ePZk75y0G7XowhR068PrUp+i9+c58+eUiAPbac1deeXkKG23Sj2+//a7532AC2tK9vFatqmCfw0/gzyP/yJB99gDgpX9N5aKRo3n1yYfoWFy83m1nfPQJvzpzGHffMppBA3YAYM8f/5zLLjyPnxw8BIAnn32RK2+4hTeez/kRGq1aXPfyOnvL4xt8Lpsw5+EWcy8vdXm1EjsP2J7OnTvxyqtT15S9/MprdO7ciZ0HbL9W3fLylRzx0xN5d/r3Nw+trg69tCUdOzJ7zn85/IhfrwkmADV/V3TsuP4Ti7ReH33yGeXlK9ll5++f9rrrwB0pL1/JR5/MqnXbsbfeycFD9lkTTAA26NqFJ597kWXLy1heVsZTz7/MDv22bbL25zt1eUmz6tW7J8uXl/FdWj/1smXLKStbQe/NNoXXv6+7YkU5Tz/z4lrbD7/gTL799jvee38m5eUreebZl9Zaf8EFZ/DJJ7PXCjKSP75a9DUlJR3p0rnTmrLOnTpR0rGYL79avN7tpn/wIW+8/R5T7p+4VvllF57HJVffyF6HHkdBQQG9ev6A+24b02Ttz3calJdmVVpawqpVFeuUr1q1iuLiolq3Pf74nzFs2JlcOuI6ystXrrP+ogvP4Zijf8LvL7witvZKy1K+ahVFWSZwFBYWUpFlYL7GQ48/yT577ErfPpuvVT533gL6brk5d958HZNuvo6uXTpz8VV/pjV3oScp1Yj/WhIFlFaivHxl1sBRXFzMihXl693u/5x0PPdOvoW//GUSE2/PfNInjLh0GNddO4Lf/f4K/vnUC7G2WVqOjsVFWQNHZWXlesdxcHjeAAANqklEQVRPKisreeHl/3DkYQetVT533gKuvWkCI/9wPoN3GcDgQTsx7to/8fpb05n2zvtN0v58V92IpSVpti4vM/sSyD7HNeLu3ZupOa3O/HkL6dy5E507d2L58jIAunTpTKdOpSyY/0XWbS44/wzG3DiS0Tfcwog/Xb/O+jE3XslvzxvKueddkjXYSP7oscnGlJevpKxsBZ06lQJh6m/5ylX02GSjrNu8+e4HVFVVsd+eu61VPtM/paiwkC232GxNWa+ePdhwg67MW7CQwYN2aro3kqdaWqbRUM2ZofwUKABGAUevZ5H1mP7eTJYvL2PffXZfU7b/fnuyfHkZ09+buU79U04+gTE3juSyy0dnDSYjr7iQ8849lVNPG6Zg0gbYNn0pKenIW9M/WFP25jvvU1LSEdu6b9Ztpn/wIf1sa0pLS9Yq/8HG3VlVUcGcufPWlC1e8g1Lv1vG5r03bZo3IK1Cs2Uo7v6GmV0EnOXuY5vruPli5cqV3HnnA9w8bhTLThtGQUEB4266mgm3TWbVqlV07dqFoqJCFi9eQo8em3DzuFE8+NDj3HX3g/Toscma/Xzzzbf0778Nl1x8PmNvuo3nn39lrfWLFy+hqqoqibcoTahjcTE//+mhXHvTBDqVlpIixbU3TeCXxxxBcXERy5aXUVlZSfcNu63Z5uPPZrNN3z7r7Gun7fvR37bhsutu4uILfkO7dgX8+ZZJbL/dNuySNhNM6q+ldV01VLNfh2JmVwIT3D17P00O2tJ1KABFRUXcNPYqTjj+Z6xeXcUDDz7GRX+4iqqqKu6cdBP777cnW2+7B0NP/SW3T7wx6z6OPOpkdt99EJdeckHW9QMGHsiMGd6Ub6PFaEvXoQBUVFRw3biJPP3Cy7Rv357DDzmAi847gw4d2jNi1BimvfMezz56z5r6J59zETvv2J/hZ5+6zr4WL/mGP99yB69NewdSKfYcPIiLLziLDbtt0JxvKXFxXYdyUp9jGnwu+9vnj7WY61B0YaO0WW0toEj84gooJzYioNzXggKKrkMREUlYS7tAsaEUUEREEqZZXiIiImmUoYiIJCxfZnkpoIiIJExjKCIiEot8GUNRQBERSZi6vEREJBat+XrAdJrlJSIisVCGIiKSMA3Ki4hILDSGIiIisdAsLxERiYW6vEREJBaa5SUiIpJGGYqISMI0KC8iIrHQoLyIiMRCg/IiIhKLfBmUV0AREUlYvmQomuUlIiKxUIYiIpIwDcqLiEgsqjWGIiIicWiucGJmA4DbgJ2AWcBQd5+Wpd4RwLXAVsBXwA3uPrGu/WsMRUQkYdWkGrzUl5kVAVOAvwPdgGuAZ82sa0a9TYFHgD+6exfgOGCcmQ2q6xgKKCIiCWuOgAIMAQrdfZy7V7r7Q8AM4IT0Su6+ENjE3Z8ys3bARsBqYFldB1CXl4hIwhpzHYqZdSNkHJmWuvvStNf9gQ8z6nwE7Ji5obsvM7NS4FtCnBjt7p/U1RZlKCIirdswYHaWZVhGvc7AioyyFUDpeva7EugE7AYMNbPT6mqIMhQRkYQ18sLGccDkLOVLM16XASUZZaXA8mw7dfdqoAJ408xuB44E7qytIQooIiIJa8x1KFG3VmbwyGYmMDyjbDvg3vQCM9sfGOvuu6QVF9fnGAooIiIJa6Z7eb0IFJjZcGA8cCxh+vDjGfXeBXqb2e+Am4HdgdOAo+s6gMZQREQS1hyzvNy9AjiMEEiWACOAo9x9kZldamYzonrfAj8Bjonq3Q6c7u4v13WMgtZ8l8sORb1bb+MlceULXk26CdLKFW7ctyCO/QzsuXeDz2XvfPHvWNoQB2UoIiISC42hiIgkLF9uX6+AIiKSMN1tWEREYqG7DYuISCyUoYiISCzyJUPRLC8REYmFMhQRkYSpy0tERGKRL11eCigiIglThiIiIrFQhiIiIrHIlwxFs7xERCQWylBERBKWSlUn3YRYKKCIiCRMN4cUEZFYtObnUqVTQBERSZgyFBERiUW+ZCia5SUiIrFQhiIikjBd2CgiIrHIlwsbFVBERBKWL2MoCigiIgnTLC8REYmFMhQREYlFvgzKa9qwiIjEQhmKiEjC1OUlIiKx0KC8iIjEQhmKiIjEIl8G5RVQREQSli9XymuWl4iIxEIZiohIwtTlJSIisdCgvIiIxCJfxlAUUEREEqYMRUREYpEvAUWzvEREJBYF+RIZRUQkWcpQREQkFgooIiISCwUUERGJhQKKiIjEQgFFRERioYAiIiKxUEAREZFYKKCIiEgsFFBERCQWupdXHjKzAcBtwE7ALGCou09LtlXS2pjZYOAf7v6DpNsirYMylDxjZkXAFODvQDfgGuBZM+uaaMOk1TCzAjM7HXgWKEq6PdJ6KKDknyFAobuPc/dKd38ImAGckGyzpBW5EjgbGJV0Q6R1UUDJP/2BDzPKPgJ2TKAt0jrd5u67AG8m3RBpXRRQ8k9nYEVG2QqgNIG2SCvk7guSboO0Tgoo+acMKMkoKwWWJ9AWEWlDFFDyz0zAMsq2i8pFRJqMpg3nnxeBAjMbDowHjiVMH3480VaJSN5ThpJn3L0COIwQSJYAI4Cj3H1Rog0TkbynRwCLiEgslKGIiEgsFFBERCQWCigiIhILBRQREYmFAoqIiMRCAUVERGKhCxsFADObA/RJK6oGlgGvARe7+/SYj/c8MM/dTzGzIYQLMjd393l1bFcAnAQ87e5f5bJtnMysHXAmcDLhTgQdgU+BB4Gx7r6yiY+/5vNryuOI5EIZiqQbDWwaLZsDBwJdCc9T6dKEx/1PdMz63JRwL+Aevr/ZZS7bxsLMOgD/BK4C7gf2BPoB1wGnA/+IAp9Im6IMRdItd/cv0l4vMLMLCSftAwkP7opddHX/F3VWDNY6Uee4bVwuJHweu7j7+2nlc8zsdcLjAn4CPNnM7RJJlAKK1GV19HMVgJmlgKuBoVH5rtG6McCRhBP+VGC4u3u0TTvgCkIXUWfgDqB9zQEyu63MrDCqfzLQHXgX+D0hcLwabTbbzK4EXsrYthS4HPgF0BOYDlzq7i9Ex5oMVBHuyvzr6PdngbPdfVlUZw7wUrbupCjzOAe4JyOYAODun5lZP+CztG32JWR/A4GFhKdpXlnTLRZ9pqcBpwC7AZ8Tus1ur8/nF9XZIfo32Jdwy50ngT+6+9K09/QI8NPoM/2xu7+d2X6RxlCXl6yXmfUFriecBP+TtuoM4AjgGOBLQvdPL+DHwD6EE+K/zGyjqP4I4ALgt8BgwgltSC2H/gvhBHseMIAQUJ4GyglBi2g/N2bZ9iHgeOAsYGdCcHvazHZPq3Mi4YS8V3SMn0ftq7Fbxut0WxG6A//f+hrv7p+6ewrAzHYGngEeIzzk7HTCSX1CxmajCTfzHEgImhPMrGZMq9bPz8x6Ay8D70Xb/5zwoLXHMo5xNiEoHUH4TEVipQxF0l1mZhdHvxdGyzvAMe7+XVq9ye7+LoCZHUQ4AXdPq3O2mf0IONPMrgfOBca4+yPRNmcCB2VrQDRWMxQ4y92nRGUXEIJJN8Jf3wCL3H25maVv259wsv6xuz8bFV8QBZMLgeOisq+B8929CnAz+xVhHASAOm6k2SP6uTij3dOBH6YV3efuv4mO+6S71wS/T83sLELAvdTdF0bld7n7w9G+LiIEnsFmNpe6P7+zgVnuflFae34BzDOzPd39taj4f9395Vrem0ijKKBIur8Ct0a/rwa+rukGyjAr7feBhL/2F6Sf3AmznvoBGxNOwm/VrHD3CjNbX3eLAUXAG2n1VxNOzJjZJrW0f4fo578zyl8l/FVe47MomNRYCvSuZb/pvo5+ds8o/ymh3RAmDXSMfh8IbGNm6Q84qxkH6kfI/gA+rlnp7kujz7KI+n1+A4GBGceo0Y8wUw/W/ncTiZ0CiqRb4u6f1qNeedrvFYSsYfcs9ZYDNbezzpz1VLGefVfW4/jrUzNVN/NY7TP2uyrLtvWdlfUZoZtvX8JYCADuPrfmdzPL/HzuIXRpZVqY9vv62lSfz6+CMA50fpZ9pGdb5VnWi8RGYyjSWDOI/lqPxg4+BWYDo4D93H0xMJ8wXgGsGWQeuJ79fUrIjnZNr29mH0fdOLU9b2FG9HPvjPK9iemJlVFmMx44NRp8X4uZFQHpWdQMoF/NZxN9PpsQxn/qnIpdz89vBiET+TztGFXAOMJ4j0izUIYijfUCYeD74Wis40vgYkIX0FVRnRuBUWb2EaEr63zCRZSvZu7M3cvM7K/ANWa2GPgEGA5sSJjNVTOGMdDMvsnY9jMze4gwoP0bYC5hEHoXYFh931DUrVbh7t+up8r1hHGjf5vZKMKgezmwR/TetwNuieqOBt42s7HA7VH7JwHzM6Zo16auz288YXLB5GjMqpjQfdmNtK40kaamDEUaJZrNdBThr+QphEH8bYFD3X1mVGccYSrvNdH6LtT+SOI/Ag8DdxFmI/UnDLR/Scg0HiXM5royy7ZnEGaE3Qe8TeiKOyRtYLo+pgE3r29lNKZzFGHm1RGEGVYzgZHR79u7+6So7vvA4YQs6d3ofb0MHF3fxtT1+UWB6SDCNOnXCQFuLnBwdJ2OSLPQExtFRCQWylBERCQWCigiIhILBRQREYmFAoqIiMRCAUVERGKhgCIiIrFQQBERkVgooIiISCwUUEREJBb/H6oee3FuLN74AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x396 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(6,5.5))\n",
    "\n",
    "\n",
    "sbn.set(font_scale=1.2)\n",
    "\n",
    "hm = sbn.heatmap(cmatrix, annot=True)\n",
    "hm.axes.set_title(\"Confusion Matrix\",fontsize=18)\n",
    "hm.set_xlabel('Prediction: Gender',fontsize=15)\n",
    "hm.set_ylabel('Gender',fontsize=15)\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "#plt.tight_layout(pad=0.4, w_pad=0.5, h_pad=1.0)\n",
    "plt.tight_layout()\n",
    "#plt.savefig('result.png')\n",
    "#plt.savefig('result.pdf')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Discussion: \n",
    "\n",
    "> The prediction is poor; especially, it is close to a random guess for the gender=0 group. We need a **better hyperparameter tuning**, or have to try **other ML algorithms**. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### (E) Describe one of your favorite tools or techniques and give a small example of how it's helped you solve a problem. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> I have built a stand-alone **Spark/Hadoop cluster** and used **Dataproc** in Google Cloud Platform (GCP) for my Big Data research. Hence, my favorite tool is `PySpark`. Though `pandas` and `scikit-learn` are powerful and handy, their capabilities are intrinsically confined in a single node machine. As a data science project, I have solved the problem of Microsoft Malware Prediction using `PySpark`. Since the size of training sample is 8 millions, it is quite challenging to solve this problem using `scikit-learn`. Many tricks to optimize memory usages and reduce training times are required to solve this problem in a single node. Though, with many fancy tricks, only limited ML algorithms and rough hyperparameter tunings are applicable. Using `PySpark`, any such single-note limitations are gone. I could solve the problem more focusing on applying various MLs with better hyperparameter tunings than struggling with optimizing `pandas` and `scikit-learn`. One tradeoff is that learning `PySpark` is harder than learning `pandas` and `scikit-learn`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}