Analyzing with BigQuery in Jupyter Notebook on GCP.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# How to analyze with BigQuery in Jupyter Notebook.\n",
    "\n",
    "This notebook shows how to carry out interactive data analysis of data in BigQuery using a Jupyter Notebook managed by Vertex AI Workbench using [flights dataset](https://www.transtats.bts.gov/DL_SelectFields.aspx?gnoyr_VQ=FGJ&QO_fu146_anzr=) in BTS website.\n",
    "\n",
    "You can find how to ingest this dataset [here](https://github.com/GoogleCloudPlatform/data-science-on-gcp/tree/main/05_bqnotebook). "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a look at our working directory."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/home/jupyter\n"
     ]
    }
   ],
   "source": [
    "!pwd"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's import the libraries we'll use in this analysis."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Magics Command in Jupyter Notebook"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can install a library using `pip` as follows:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: pytz in /opt/conda/lib/python3.7/site-packages (2022.2.1)\n",
      "Note: you may need to restart the kernel to use updated packages.\n"
     ]
    }
   ],
   "source": [
    "%pip install pytz"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can use another language magic command like %%. Let's, for example, type a html code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "This is a <b> HTML </b> string.\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%%html\n",
    "This is a <b> HTML </b> string."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Using BigQuery commands in Notebook."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can convert SQL results into a Pandas DataFrame using `%%bigquery` as follows:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 1/1 [00:00<00:00, 909.43query/s] \n",
      "Downloading: 100%|██████████| 5/5 [00:03<00:00,  1.46rows/s]\n"
     ]
    },
    {
     "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>ORIGIN</th>\n",
       "      <th>dep_delay</th>\n",
       "      <th>arr_delay</th>\n",
       "      <th>num_flights</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>LFT</td>\n",
       "      <td>7.627476</td>\n",
       "      <td>7.126566</td>\n",
       "      <td>5428</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>FCA</td>\n",
       "      <td>1.385915</td>\n",
       "      <td>-2.739028</td>\n",
       "      <td>2119</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>MBS</td>\n",
       "      <td>7.557350</td>\n",
       "      <td>4.908584</td>\n",
       "      <td>1794</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>LIT</td>\n",
       "      <td>9.049897</td>\n",
       "      <td>5.260500</td>\n",
       "      <td>11167</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>DTW</td>\n",
       "      <td>8.794979</td>\n",
       "      <td>3.216070</td>\n",
       "      <td>116874</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  ORIGIN  dep_delay  arr_delay  num_flights\n",
       "0    LFT   7.627476   7.126566         5428\n",
       "1    FCA   1.385915  -2.739028         2119\n",
       "2    MBS   7.557350   4.908584         1794\n",
       "3    LIT   9.049897   5.260500        11167\n",
       "4    DTW   8.794979   3.216070       116874"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%bigquery\n",
    "SELECT \n",
    "   ORIGIN,\n",
    "   AVG(DEP_DELAY) AS dep_delay,\n",
    "   AVG(ARR_DELAY) AS arr_delay,\n",
    "   COUNT(ARR_DELAY) AS num_flights\n",
    " FROM\n",
    "   dsongcp.flights_tzcorr\n",
    " GROUP BY\n",
    "   ORIGIN\n",
    "LIMIT 5"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can also directly query BigQuery with the Python library:"
   ]
  },
  {
   "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>ORIGIN</th>\n",
       "      <th>dep_delay</th>\n",
       "      <th>arr_delay</th>\n",
       "      <th>num_flights</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>LFT</td>\n",
       "      <td>7.627476</td>\n",
       "      <td>7.126566</td>\n",
       "      <td>5428</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>FCA</td>\n",
       "      <td>1.385915</td>\n",
       "      <td>-2.739028</td>\n",
       "      <td>2119</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>MBS</td>\n",
       "      <td>7.557350</td>\n",
       "      <td>4.908584</td>\n",
       "      <td>1794</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>LIT</td>\n",
       "      <td>9.049897</td>\n",
       "      <td>5.260500</td>\n",
       "      <td>11167</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>DTW</td>\n",
       "      <td>8.794979</td>\n",
       "      <td>3.216070</td>\n",
       "      <td>116874</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  ORIGIN  dep_delay  arr_delay  num_flights\n",
       "0    LFT   7.627476   7.126566         5428\n",
       "1    FCA   1.385915  -2.739028         2119\n",
       "2    MBS   7.557350   4.908584         1794\n",
       "3    LIT   9.049897   5.260500        11167\n",
       "4    DTW   8.794979   3.216070       116874"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from google.cloud import bigquery\n",
    "\n",
    "# Write a SQL query \n",
    "sql = \"\"\"\n",
    "SELECT \n",
    "   ORIGIN,\n",
    "   AVG(DEP_DELAY) AS dep_delay,\n",
    "   AVG(ARR_DELAY) AS arr_delay,\n",
    "   COUNT(ARR_DELAY) AS num_flights\n",
    " FROM\n",
    "   dsongcp.flights_tzcorr\n",
    " GROUP BY\n",
    "   ORIGIN\n",
    "LIMIT 5\n",
    "\"\"\"\n",
    "\n",
    "# Clienting to bundle configuration needed for API requests.\n",
    "bq = bigquery.Client()\n",
    "\n",
    "# Querying the sql command and then converting into Pandas DataFrame\n",
    "bq.query(sql).to_dataframe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I'm going to plot a Probability Distribution Function (PDF) of different arrival delays. We can assign the output to a variable, df."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 1/1 [00:00<00:00, 340.42query/s] \n",
      "Downloading: 100%|██████████| 1294778/1294778 [00:02<00:00, 537033.80rows/s] \n"
     ]
    }
   ],
   "source": [
    "%%bigquery df\n",
    "SELECT ARR_DELAY, DEP_DELAY\n",
    "FROM dsongcp.flights_tzcorr\n",
    "WHERE DEP_DELAY >= 10"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a look at the type of the variable."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "pandas.core.frame.DataFrame"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "type(df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can use Pandas methods you want."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "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>ARR_DELAY</th>\n",
       "      <th>DEP_DELAY</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1.286778e+06</td>\n",
       "      <td>1.294778e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>4.611797e+01</td>\n",
       "      <td>5.094516e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>6.360700e+01</td>\n",
       "      <td>6.151423e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>-7.800000e+01</td>\n",
       "      <td>1.000000e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>1.100000e+01</td>\n",
       "      <td>1.700000e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>2.700000e+01</td>\n",
       "      <td>3.000000e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>5.900000e+01</td>\n",
       "      <td>6.100000e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.971000e+03</td>\n",
       "      <td>1.988000e+03</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          ARR_DELAY     DEP_DELAY\n",
       "count  1.286778e+06  1.294778e+06\n",
       "mean   4.611797e+01  5.094516e+01\n",
       "std    6.360700e+01  6.151423e+01\n",
       "min   -7.800000e+01  1.000000e+01\n",
       "25%    1.100000e+01  1.700000e+01\n",
       "50%    2.700000e+01  3.000000e+01\n",
       "75%    5.900000e+01  6.100000e+01\n",
       "max    1.971000e+03  1.988000e+03"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Visualization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's draw a violinplot for the `ARR_DELAY` variable."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.set_style(\"darkgrid\")\n",
    "# sns.set(font_scale = 1.5)\n",
    "ax = sns.violinplot(data=df, x='ARR_DELAY', inner='box', orient='h')\n",
    "ax.axes.set_xlim(-50, 300);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Plotting distributions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 1/1 [00:00<00:00, 484.78query/s] \n",
      "Downloading: 100%|██████████| 5819079/5819079 [00:04<00:00, 1267678.87rows/s]\n"
     ]
    }
   ],
   "source": [
    "%%bigquery df\n",
    "SELECT ARR_DELAY, DEP_DELAY\n",
    "FROM dsongcp.flights_tzcorr"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's create a variable that shows a delay of less than 10 minutes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "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>ARR_DELAY</th>\n",
       "      <th>DEP_DELAY</th>\n",
       "      <th>ontime</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>264.0</td>\n",
       "      <td>245.0</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>54.0</td>\n",
       "      <td>54.0</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>128.0</td>\n",
       "      <td>145.0</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>40.0</td>\n",
       "      <td>39.0</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>29.0</td>\n",
       "      <td>23.0</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   ARR_DELAY  DEP_DELAY  ontime\n",
       "1      264.0      245.0   False\n",
       "5       54.0       54.0   False\n",
       "6      128.0      145.0   False\n",
       "7       40.0       39.0   False\n",
       "8       29.0       23.0   False"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['ontime'] = df['DEP_DELAY'] < 10\n",
    "df[df['ARR_DELAY'] > 0].head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's draw the violinplot using the `ontime` variable."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.set_style(\"whitegrid\")\n",
    "ax = sns.violinplot(data=df, x='ARR_DELAY', y='ontime', inner='box', orient='h')\n",
    "ax.set_xlim(-50, 200);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Finding Outlier values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 1/1 [00:00<00:00, 795.88query/s] \n",
      "Downloading: 100%|██████████| 1218/1218 [00:03<00:00, 372.02rows/s]\n"
     ]
    }
   ],
   "source": [
    "%%bigquery depdelay\n",
    "SELECT\n",
    "  DEP_DELAY,\n",
    "  AVG(ARR_DELAY) AS arrival_delay,\n",
    "  COUNT(ARR_DELAY) AS numflights\n",
    "FROM\n",
    "  dsongcp.flights_tzcorr\n",
    "GROUP BY\n",
    "  DEP_DELAY\n",
    "ORDER BY\n",
    "  DEP_DELAY"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1218, 3)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "depdelay.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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>DEP_DELAY</th>\n",
       "      <th>arrival_delay</th>\n",
       "      <th>numflights</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-82.0</td>\n",
       "      <td>-80.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-68.0</td>\n",
       "      <td>-87.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-61.0</td>\n",
       "      <td>-77.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-56.0</td>\n",
       "      <td>-26.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   DEP_DELAY  arrival_delay  numflights\n",
       "0        NaN            NaN           0\n",
       "1      -82.0          -80.0           1\n",
       "2      -68.0          -87.0           1\n",
       "3      -61.0          -77.0           1\n",
       "4      -56.0          -26.0           1"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "depdelay.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Filtering Data on Occurence Frequency"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 1/1 [00:00<00:00, 765.52query/s] \n",
      "Downloading: 100%|██████████| 9/9 [00:03<00:00,  2.90rows/s]\n"
     ]
    }
   ],
   "source": [
    "%%bigquery df\n",
    "DECLARE total_flights INT64;\n",
    "SET total_flights = (\n",
    "    SELECT COUNT(*) FROM dsongcp.flights_tzcorr\n",
    ");\n",
    "\n",
    "CREATE TEMPORARY FUNCTION linear_fit(NUM_TOTAL INT64, THRESH INT64)\n",
    "RETURNS STRUCT<thresh INT64, num_removed INT64, lm FLOAT64>\n",
    "AS ((\n",
    "    SELECT AS STRUCT\n",
    "        THRESH,\n",
    "        (NUM_TOTAL - SUM(numflights)) AS num_removed,\n",
    "        ROUND(AVG(arrival_delay * numflights) / AVG(dep_delay * numflights), 2) AS lm\n",
    "    FROM\n",
    "    (\n",
    "        SELECT\n",
    "            DEP_DELAY,\n",
    "            AVG(ARR_DELAY) AS arrival_delay,\n",
    "            STDDEV(ARR_DELAY) AS stddev_arrival_delay,\n",
    "            COUNT(ARR_DELAY) AS numflights\n",
    "        FROM\n",
    "            dsongcp.flights_tzcorr\n",
    "        GROUP BY\n",
    "            DEP_DELAY\n",
    "    )\n",
    "    WHERE numflights > THRESH\n",
    "));\n",
    "\n",
    "SELECT linear_fit(total_flights, 1000) stats\n",
    "UNION ALL SELECT linear_fit(total_flights, 500)\n",
    "UNION ALL SELECT linear_fit(total_flights, 370)\n",
    "UNION ALL SELECT linear_fit(total_flights, 300)\n",
    "UNION ALL SELECT linear_fit(total_flights, 200)\n",
    "UNION ALL SELECT linear_fit(total_flights, 100)\n",
    "UNION ALL SELECT linear_fit(total_flights, 22)\n",
    "UNION ALL SELECT linear_fit(total_flights, 10)\n",
    "UNION ALL SELECT linear_fit(total_flights, 5)\n",
    "ORDER BY stats.thresh DESC"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a look at the thresh and the number of removed values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    (1000, 175873, 0.25)\n",
       "1     (500, 143801, 0.34)\n",
       "2     (370, 135518, 0.36)\n",
       "3     (300, 129835, 0.38)\n",
       "4      (200, 123640, 0.4)\n",
       "5     (100, 115471, 0.43)\n",
       "6      (22, 108247, 0.45)\n",
       "7      (10, 106958, 0.46)\n",
       "8       (5, 106319, 0.46)\n",
       "Name: stats, dtype: object"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['stats'].map(lambda x: (x['thresh'], x['num_removed'], x['lm']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Arrival delay conditioned on departure delay"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 1/1 [00:00<00:00, 699.75query/s] \n",
      "Downloading: 100%|██████████| 240/240 [00:05<00:00, 44.30rows/s]\n"
     ]
    }
   ],
   "source": [
    "%%bigquery depdelay\n",
    "SELECT\n",
    "    DEP_DELAY,\n",
    "    AVG(ARR_DELAY) AS arrival_delay,\n",
    "    STDDEV(ARR_DELAY) AS stddev_arrival_delay,\n",
    "    COUNT(ARR_DELAY) AS numflights\n",
    "FROM\n",
    "    dsongcp.flights_tzcorr\n",
    "GROUP BY\n",
    "    DEP_DELAY\n",
    "HAVING numflights > 370\n",
    "ORDER BY DEP_DELAY"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "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>DEP_DELAY</th>\n",
       "      <th>arrival_delay</th>\n",
       "      <th>stddev_arrival_delay</th>\n",
       "      <th>numflights</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-23.0</td>\n",
       "      <td>-23.888646</td>\n",
       "      <td>11.432163</td>\n",
       "      <td>458</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-22.0</td>\n",
       "      <td>-23.227488</td>\n",
       "      <td>12.590133</td>\n",
       "      <td>633</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-21.0</td>\n",
       "      <td>-22.299781</td>\n",
       "      <td>11.558313</td>\n",
       "      <td>914</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-20.0</td>\n",
       "      <td>-21.407821</td>\n",
       "      <td>12.066489</td>\n",
       "      <td>1432</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-19.0</td>\n",
       "      <td>-20.430769</td>\n",
       "      <td>11.910134</td>\n",
       "      <td>1950</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   DEP_DELAY  arrival_delay  stddev_arrival_delay  numflights\n",
       "0      -23.0     -23.888646             11.432163         458\n",
       "1      -22.0     -23.227488             12.590133         633\n",
       "2      -21.0     -22.299781             11.558313         914\n",
       "3      -20.0     -21.407821             12.066489        1432\n",
       "4      -19.0     -20.430769             11.910134        1950"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "depdelay.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = depdelay.plot(kind='line', x='DEP_DELAY', \n",
    "              y='arrival_delay', yerr='stddev_arrival_delay')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Applying a probabilistic decision threshold"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that our decision criteria are 15 minutes and 30%. Let's find a threshold with the assumption that it is normally distributed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "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>DEP_DELAY</th>\n",
       "      <th>arrival_delay</th>\n",
       "      <th>stddev_arrival_delay</th>\n",
       "      <th>numflights</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-23.0</td>\n",
       "      <td>-23.888646</td>\n",
       "      <td>11.432163</td>\n",
       "      <td>458</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-22.0</td>\n",
       "      <td>-23.227488</td>\n",
       "      <td>12.590133</td>\n",
       "      <td>633</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-21.0</td>\n",
       "      <td>-22.299781</td>\n",
       "      <td>11.558313</td>\n",
       "      <td>914</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-20.0</td>\n",
       "      <td>-21.407821</td>\n",
       "      <td>12.066489</td>\n",
       "      <td>1432</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-19.0</td>\n",
       "      <td>-20.430769</td>\n",
       "      <td>11.910134</td>\n",
       "      <td>1950</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   DEP_DELAY  arrival_delay  stddev_arrival_delay  numflights\n",
       "0      -23.0     -23.888646             11.432163         458\n",
       "1      -22.0     -23.227488             12.590133         633\n",
       "2      -21.0     -22.299781             11.558313         914\n",
       "3      -20.0     -21.407821             12.066489        1432\n",
       "4      -19.0     -20.430769             11.910134        1950"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "depdelay.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Obtaining Z-value for 30% likely.\n",
    "Z_30 = 0.52\n",
    "\n",
    "# Caluculating Z-score\n",
    "depdelay['arr_delay_30'] = (Z_30 * depdelay['stddev_arrival_delay']) \\\n",
    "             + depdelay['arrival_delay']\n",
    "\n",
    "# Plotting arrival values\n",
    "ax = plt.axes()\n",
    "depdelay.plot(kind='line', x='DEP_DELAY', y='arr_delay_30',\n",
    "              ax=ax, ylim=(0,30), xlim=(0,30), legend=False)\n",
    "ax.set_xlabel('Departure Delay (minutes)')\n",
    "ax.set_ylabel('> 30% prob of this\\n Arrival Delay (minutes)');\n",
    "\n",
    "x = np.arange(0, 30)\n",
    "y = np.ones_like(x) * 15\n",
    "ax.plot(x, y, 'r.');\n",
    "\n",
    "y = np.arange(0, 30)\n",
    "x = np.ones_like(y) * 13\n",
    "ax.plot(x, y, 'g.');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As you can see from graph, our decision criteria translate to a departure delay of 13 minutes. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Empirical probability distribution function"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's find a threshold without the assumption that it is normally distributed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 3/3 [00:00<00:00, 1136.15query/s]                        \n",
      "Downloading: 100%|██████████| 240/240 [00:05<00:00, 44.35rows/s]\n"
     ]
    }
   ],
   "source": [
    "%%bigquery depdelay\n",
    "SELECT\n",
    "    DEP_DELAY,\n",
    "    APPROX_QUANTILES(ARR_DELAY, 101)[OFFSET(70)] AS arrival_delay,\n",
    "    COUNT(ARR_DELAY) AS numflights\n",
    "FROM\n",
    "    dsongcp.flights_tzcorr\n",
    "GROUP BY\n",
    "    DEP_DELAY\n",
    "HAVING numflights > 370\n",
    "ORDER BY DEP_DELAY"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = plt.axes()\n",
    "depdelay.plot(kind='line', x='DEP_DELAY', y='arrival_delay',\n",
    "              ax=ax, ylim=(0,30), xlim=(0,30), legend=False)\n",
    "ax.set_xlabel('Departure Delay (minutes)')\n",
    "ax.set_ylabel('> 30% prob of this\\n Arrival Delay (minutes)');\n",
    "\n",
    "x = np.arange(0, 30)\n",
    "y = np.ones_like(x) * 15\n",
    "ax.plot(x, y, 'r.');\n",
    "\n",
    "y = np.arange(0, 30)\n",
    "x = np.ones_like(y) * 16\n",
    "ax.plot(x, y, 'g.');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As you can see from this graph, our decision threshold, without the assumption of normal distribution, is 16 minutes."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Creating training/evaluation dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a look at all the unique days in the dataset:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 1/1 [00:00<00:00, 715.75query/s] \n",
      "Downloading: 100%|██████████| 10/10 [00:03<00:00,  3.07rows/s]\n"
     ]
    },
    {
     "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>FL_DATE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2015-01-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2015-01-02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2015-01-03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2015-01-04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2015-01-05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2015-01-06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2015-01-07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2015-01-08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2015-01-09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>2015-01-10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      FL_DATE\n",
       "0  2015-01-01\n",
       "1  2015-01-02\n",
       "2  2015-01-03\n",
       "3  2015-01-04\n",
       "4  2015-01-05\n",
       "5  2015-01-06\n",
       "6  2015-01-07\n",
       "7  2015-01-08\n",
       "8  2015-01-09\n",
       "9  2015-01-10"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%bigquery\n",
    "SELECT\n",
    "  DISTINCT(FL_DATE) AS FL_DATE\n",
    "FROM\n",
    "  dsongcp.flights_tzcorr\n",
    "ORDER BY\n",
    "  FL_DATE\n",
    "LIMIT 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 1/1 [00:00<00:00, 903.56query/s] \n",
      "Downloading: 100%|██████████| 5/5 [00:03<00:00,  1.51rows/s]\n"
     ]
    },
    {
     "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>FL_DATE</th>\n",
       "      <th>is_train_day</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2015-01-01</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2015-01-02</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2015-01-03</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2015-01-04</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2015-01-05</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      FL_DATE is_train_day\n",
       "0  2015-01-01         True\n",
       "1  2015-01-02        False\n",
       "2  2015-01-03        False\n",
       "3  2015-01-04         True\n",
       "4  2015-01-05         True"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%bigquery\n",
    "SELECT\n",
    "  FL_DATE,\n",
    "  IF(ABS(MOD(FARM_FINGERPRINT(CAST(FL_DATE AS STRING)), 100)) < 70,\n",
    "     'True', 'False') AS is_train_day\n",
    "FROM (\n",
    "  SELECT\n",
    "    DISTINCT(FL_DATE) AS FL_DATE\n",
    "  FROM\n",
    "    dsongcp.flights_tzcorr)\n",
    "ORDER BY\n",
    "  FL_DATE\n",
    "LIMIT 5"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a look at the number of categories of train day."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 1/1 [00:00<00:00, 800.75query/s] \n",
      "Downloading: 100%|██████████| 2/2 [00:05<00:00,  2.58s/rows]\n"
     ]
    },
    {
     "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>f0_</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>263</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>102</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   f0_\n",
       "0  263\n",
       "1  102"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%bigquery\n",
    "WITH train_day AS (\n",
    "SELECT\n",
    "  FL_DATE,\n",
    "  IF(ABS(MOD(FARM_FINGERPRINT(CAST(FL_DATE AS STRING)), 100)) < 70,\n",
    "     'True', 'False') AS is_train_day\n",
    "FROM (\n",
    "  SELECT\n",
    "    DISTINCT(FL_DATE) AS FL_DATE\n",
    "  FROM\n",
    "    dsongcp.flights_tzcorr)\n",
    "ORDER BY\n",
    "  FL_DATE\n",
    ")\n",
    "\n",
    "SELECT\n",
    "    COUNT(*) \n",
    "FROM train_day\n",
    "\n",
    "GROUP BY\n",
    "    is_train_day"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's create the training table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 3/3 [00:00<00:00, 741.04query/s]                         \n"
     ]
    },
    {
     "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",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: []\n",
       "Index: []"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%bigquery\n",
    "CREATE OR REPLACE TABLE dsongcp.trainday AS\n",
    "\n",
    "SELECT\n",
    "  FL_DATE,\n",
    "  IF(ABS(MOD(FARM_FINGERPRINT(CAST(FL_DATE AS STRING)), 100)) < 70,\n",
    "     'True', 'False') AS is_train_day\n",
    "FROM (\n",
    "  SELECT\n",
    "    DISTINCT(FL_DATE) AS FL_DATE\n",
    "  FROM\n",
    "    dsongcp.flights_tzcorr)\n",
    "ORDER BY\n",
    "  FL_DATE"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's assign the training data to the depdelay variable."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 4/4 [00:00<00:00, 1900.24query/s]                        \n",
      "Downloading: 100%|██████████| 224/224 [00:01<00:00, 116.41rows/s]\n"
     ]
    }
   ],
   "source": [
    "%%bigquery depdelay\n",
    "SELECT\n",
    "    DEP_DELAY,\n",
    "    APPROX_QUANTILES(ARR_DELAY, 101)[OFFSET(70)] AS arrival_delay,\n",
    "    COUNT(ARR_DELAY) AS numflights\n",
    "FROM\n",
    "    dsongcp.flights_tzcorr\n",
    "JOIN dsongcp.trainday USING(FL_DATE)\n",
    "WHERE is_train_day = 'True'\n",
    "GROUP BY\n",
    "    DEP_DELAY\n",
    "HAVING numflights > 370\n",
    "ORDER BY DEP_DELAY"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = plt.axes()\n",
    "depdelay.plot(kind='line', x='DEP_DELAY', y='arrival_delay',\n",
    "              ax=ax, ylim=(0,30), xlim=(0,30), legend=False)\n",
    "ax.set_xlabel('Departure Delay (minutes)')\n",
    "ax.set_ylabel('> 30% prob of this\\n Arrival Delay (minutes)');\n",
    "\n",
    "x = np.arange(0, 30)\n",
    "y = np.ones_like(x) * 15\n",
    "ax.plot(x, y, 'r.');\n",
    "\n",
    "y = np.arange(0, 30)\n",
    "x = np.ones_like(y) * 16\n",
    "ax.plot(x, y, 'g.');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We obtain the same answer—16 minutes—after creating the empirical probabilistic model on just the training data. Now let’s formally evaluate how well our recommendation of 16 minutes does in terms of predicting an arrival delay of 15 minutes or more. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 3/3 [00:00<00:00, 1312.63query/s]                        \n",
      "Downloading: 100%|██████████| 1/1 [00:02<00:00,  2.54s/rows]\n"
     ]
    }
   ],
   "source": [
    "%%bigquery df_eval\n",
    "SELECT\n",
    "  SUM(IF(DEP_DELAY < 16\n",
    "      AND arr_delay < 15, 1, 0)) AS correct_nocancel,\n",
    "  SUM(IF(DEP_DELAY < 16\n",
    "      AND arr_delay >= 15, 1, 0)) AS wrong_nocancel,\n",
    "  SUM(IF(DEP_DELAY >= 16\n",
    "      AND arr_delay < 15, 1, 0)) AS wrong_cancel,\n",
    "  SUM(IF(DEP_DELAY >= 16\n",
    "      AND arr_delay >= 15, 1, 0)) AS correct_cancel\n",
    "FROM (\n",
    "  SELECT\n",
    "    DEP_DELAY,\n",
    "    ARR_DELAY\n",
    "  FROM\n",
    "    dsongcp.flights_tzcorr\n",
    "  JOIN dsongcp.trainday USING(FL_DATE)\n",
    "  WHERE is_train_day = 'False' \n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a look at the results we got."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0    0.950158\n",
      "dtype: float64\n",
      "0    0.804703\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "print(df_eval['correct_nocancel'] /\n",
    "      (df_eval['correct_nocancel'] + df_eval['wrong_nocancel']))\n",
    "print(df_eval['correct_cancel'] / \n",
    "      (df_eval['correct_cancel'] + df_eval['wrong_cancel']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It turns out when you not cancel your meeting, you will be correct 95% of the time, and when you cancel your meeting, you will be correct 82% of the time."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "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>correct_nocancel</th>\n",
       "      <th>wrong_nocancel</th>\n",
       "      <th>wrong_cancel</th>\n",
       "      <th>correct_cancel</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1259740</td>\n",
       "      <td>66081</td>\n",
       "      <td>52827</td>\n",
       "      <td>217669</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   correct_nocancel  wrong_nocancel  wrong_cancel  correct_cancel\n",
       "0           1259740           66081         52827          217669"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_eval.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's evaluate right at the decision boundary by changing our scoring function with test dataset:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Query complete after 0.00s: 100%|██████████| 3/3 [00:00<00:00, 1158.11query/s]                        \n",
      "Downloading: 100%|██████████| 1/1 [00:01<00:00,  1.83s/rows]\n"
     ]
    }
   ],
   "source": [
    "%%bigquery df_eval\n",
    "SELECT\n",
    "  SUM(IF(DEP_DELAY = 15\n",
    "      AND arr_delay < 15, 1, 0)) AS correct_nocancel,\n",
    "  SUM(IF(DEP_DELAY = 15\n",
    "      AND arr_delay >= 15, 1, 0)) AS wrong_nocancel,\n",
    "  SUM(IF(DEP_DELAY = 16\n",
    "      AND arr_delay < 15, 1, 0)) AS wrong_cancel,\n",
    "  SUM(IF(DEP_DELAY = 16\n",
    "      AND arr_delay >= 15, 1, 0)) AS correct_cancel\n",
    "FROM (\n",
    "  SELECT\n",
    "    DEP_DELAY,\n",
    "    ARR_DELAY\n",
    "  FROM\n",
    "    dsongcp.flights_tzcorr\n",
    "  JOIN dsongcp.trainday USING(FL_DATE)\n",
    "  WHERE is_train_day = 'False' \n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "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>correct_nocancel</th>\n",
       "      <th>wrong_nocancel</th>\n",
       "      <th>wrong_cancel</th>\n",
       "      <th>correct_cancel</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>7684</td>\n",
       "      <td>2935</td>\n",
       "      <td>6787</td>\n",
       "      <td>2942</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   correct_nocancel  wrong_nocancel  wrong_cancel  correct_cancel\n",
       "0              7684            2935          6787            2942"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_eval.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0    0.723609\n",
      "dtype: float64\n",
      "0    0.302395\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "print(df_eval['correct_nocancel'] / (df_eval['correct_nocancel'] + df_eval['wrong_nocancel']))\n",
    "print(df_eval['correct_cancel'] / (df_eval['correct_cancel'] + df_eval['wrong_cancel']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As expected, we are correct to not cancel the meeting 72% of the time, close to our target of 70%."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Resources\n",
    "\n",
    "- [Data science on google cloud](https://www.amazon.it/Data-Science-Google-Cloud-Platform-dp-1098118952/dp/1098118952/ref=dp_ob_title_bk)"
   ]
  }
 ],
 "metadata": {
  "environment": {
   "kernel": "python3",
   "name": "common-cpu.m97",
   "type": "gcloud",
   "uri": "gcr.io/deeplearning-platform-release/base-cpu:m97"
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}