Flights_RFC_Model.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "34802e77",
   "metadata": {},
   "source": [
    "## Import dependencies used in entire notebook"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "2e8096e7",
   "metadata": {},
   "outputs": [],
   "source": [
    "import dask.dataframe as dd\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "from pathlib import Path"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f078cbb1",
   "metadata": {},
   "source": [
    "## Connect to database and import data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "90130edd",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import engine to make connection to database\n",
    "\n",
    "from sqlalchemy import create_engine"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "17b6b2cb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Connect to Flights_DB\n",
    "\n",
    "db_path = Path('Resources/Flights_DB.sqlite')\n",
    "engine = create_engine(f'sqlite:///{db_path}')\n",
    "\n",
    "conn = engine.connect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "055fe513",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load Flights table, join Cancelled_Flights and Delayed_Flights for reasons\n",
    "\n",
    "flights_df = pd.read_sql_query('''\n",
    "                                  SELECT F.Operating_Airline\n",
    "                                      ,F.Origin_Airport\n",
    "                                      ,F.Destination_Airport\n",
    "                                      ,F.Flight_Date\n",
    "                                      ,F.Scheduled_Arrival_Time\n",
    "                                      ,F.Scheduled_Departure_Time\n",
    "                                      ,D.Carrier_Delay_Minutes\n",
    "                                      ,D.Weather_Delay_Minutes\n",
    "                                      ,D.NAS_Delay_Minutes\n",
    "                                      ,D.Security_Delay_Minutes\n",
    "                                      ,D.Late_Aircraft_Delay_Minutes\n",
    "                                      ,C.Cancellation_Code\n",
    "                                      ,F.Target\n",
    "                                  FROM Flights F\n",
    "                                  LEFT JOIN Cancelled_Flights C ON F.ID = C.ID\n",
    "                                  LEFT JOIN Delayed_Flights D ON F.ID = D.ID\n",
    "                               ''',\n",
    "                               con=engine,\n",
    "                               dtype={\n",
    "                                   'Operating_Airline': object,\n",
    "                                   'Origin_Airport': object,\n",
    "                                   'Destination_Airport': object,\n",
    "                                   'Flight_Date': np.datetime64,\n",
    "                                   'Scheduled_Arrival_Time': np.datetime64,\n",
    "                                   'Scheduled_Departure_Time': np.datetime64,\n",
    "                                   'Carrier_Delay_Minutes': object,\n",
    "                                   'Weather_Delay_Minutes': object,\n",
    "                                   'NAS_Delay_Minutes': object,\n",
    "                                   'Security_Delay_Minutes': object,\n",
    "                                   'Late_Aircraft_Delay_Minutes': object,\n",
    "                                   'Cancellation_Code': object,\n",
    "                                   'Target': np.int32\n",
    "                               })\n",
    "\n",
    "flights_df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fbf7ec23",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Checking to make sure timedeltas have values\n",
    "\n",
    "flights_df.nunique()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "263c072e",
   "metadata": {},
   "source": [
    "## Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c9bfd991",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Make separate columns for Month, Day and Weekday\n",
    "\n",
    "flights_df['Month'] = flights_df['Flight_Date'].dt.month\n",
    "flights_df['Day'] = flights_df['Flight_Date'].dt.day\n",
    "flights_df['Weekday'] = flights_df['Flight_Date'].dt.dayofweek\n",
    "\n",
    "flights_df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "156e6cd8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Extract hour from arrival/departure columns\n",
    "\n",
    "time_columns = ['Scheduled_Arrival_Time', 'Scheduled_Departure_Time']\n",
    "\n",
    "for col in time_columns:\n",
    "    flights_df[col] = flights_df[col].dt.hour"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6e4ab598",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Fill None in intervals with 0\n",
    "\n",
    "none_cols = ['Carrier_Delay_Minutes', 'Weather_Delay_Minutes', 'NAS_Delay_Minutes',\n",
    "                 'Security_Delay_Minutes', 'Late_Aircraft_Delay_Minutes', 'Cancellation_Code']\n",
    "\n",
    "flights_df[none_cols] = flights_df[none_cols].apply(lambda x: x.fillna(0))\n",
    "\n",
    "flights_df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "03028697",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Double check existence of nulls\n",
    "\n",
    "flights_df.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "27707285",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Drop Flight_Date column\n",
    "\n",
    "flights_df = flights_df.drop(columns='Flight_Date')\n",
    "\n",
    "flights_df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "00ce7840",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Extract numbers from Minutes column string\n",
    "\n",
    "interval_cols = ['Carrier_Delay_Minutes', 'Weather_Delay_Minutes', 'NAS_Delay_Minutes',\n",
    "                 'Security_Delay_Minutes', 'Late_Aircraft_Delay_Minutes']\n",
    "\n",
    "flights_df[interval_cols] = flights_df[interval_cols].apply(lambda x: pd.to_timedelta(x).dt.total_seconds().astype(int) // 60)\n",
    "\n",
    "flights_df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "13c4ab21",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Check unique values to make sure no data lost\n",
    "\n",
    "flights_df.nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c910e7a1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Make columns that track reasons for both delays and cancellations as a binary\n",
    "\n",
    "def check_condition(delay_col, cancel_code):\n",
    "    \n",
    "    condition = ((flights_df[delay_col] > 0) | (flights_df['Cancellation_Code'] == cancel_code))\n",
    "    \n",
    "    return condition\n",
    "\n",
    "carrier = check_condition('Carrier_Delay_Minutes', 'A')\n",
    "bad_weather = check_condition('Weather_Delay_Minutes', 'B')\n",
    "NAS = check_condition('NAS_Delay_Minutes', 'C')\n",
    "security = check_condition('Security_Delay_Minutes', 'D')\n",
    "late_aircraft = (flights_df['Late_Aircraft_Delay_Minutes'] > 0)\n",
    "\n",
    "flights_df['Carrier_Issue'] = np.where(carrier, 1, 0)\n",
    "flights_df['Bad_Weather'] = np.where(bad_weather, 1, 0)\n",
    "flights_df['NAS'] = np.where(NAS, 1, 0)\n",
    "flights_df['Security_Issue'] = np.where(security, 1, 0)\n",
    "flights_df['Late_Aircraft'] = np.where(security, 1, 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "671252d0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create function to output encoding dictionaries \n",
    "\n",
    "def encoder (MasterDF, ColName, Dictionary):\n",
    "    uni_row_vals =[MasterDF[ColName].unique()][0]\n",
    "    for x in range(len(uni_row_vals)):\n",
    "        Dictionary[uni_row_vals[x]] = x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cacfc26f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Integer encode Origin and Destination airports\n",
    "\n",
    "loc_dict = {}\n",
    "encoder(flights_df, 'Origin_Airport', loc_dict)\n",
    "flights_df['Origin_Airport'] = flights_df['Origin_Airport'].apply(lambda x: loc_dict[x])\n",
    "flights_df['Destination_Airport'] = flights_df['Destination_Airport'].apply(lambda x: loc_dict[x])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "92cd760f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Integer encode operating airlines\n",
    "\n",
    "airline_dict = {}\n",
    "encoder(flights_df, 'Operating_Airline', airline_dict)\n",
    "flights_df['Operating_Airline'] = flights_df['Operating_Airline'].apply(lambda x: airline_dict[x])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1cff4649",
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "# Drop extra columns and check for complete DF\n",
    "\n",
    "drop_cols = ['Carrier_Delay_Minutes', 'Weather_Delay_Minutes', 'NAS_Delay_Minutes',\n",
    "                 'Security_Delay_Minutes', 'Late_Aircraft_Delay_Minutes', 'Cancellation_Code']\n",
    "\n",
    "flights_df = flights_df.drop(columns=drop_cols).copy()\n",
    "\n",
    "flights_df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d12d16c8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Checkpoint DF for use in future sessions\n",
    "\n",
    "# model_df_path = Path('Resources/flights_model_df.csv')\n",
    "# flights_df.to_csv(model_df_path, index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "595a5161",
   "metadata": {},
   "source": [
    "## Load DF as Dask DF and start parallel processing client"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "1d91eda2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import client from dask\n",
    "\n",
    "from dask.distributed import Client\n",
    "\n",
    "# Import necessary functions for machine learning models\n",
    "\n",
    "from dask_ml.model_selection import train_test_split\n",
    "from imblearn.ensemble import BalancedRandomForestClassifier\n",
    "from imblearn.metrics import classification_report_imbalanced\n",
    "from joblib import parallel_backend\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.metrics import accuracy_score, balanced_accuracy_score, classification_report, confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "acef46c6",
   "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>Operating_Airline</th>\n",
       "      <th>Origin_Airport</th>\n",
       "      <th>Destination_Airport</th>\n",
       "      <th>Scheduled_Arrival_Time</th>\n",
       "      <th>Scheduled_Departure_Time</th>\n",
       "      <th>Target</th>\n",
       "      <th>Month</th>\n",
       "      <th>Day</th>\n",
       "      <th>Weekday</th>\n",
       "      <th>Carrier_Issue</th>\n",
       "      <th>Bad_Weather</th>\n",
       "      <th>NAS</th>\n",
       "      <th>Security_Issue</th>\n",
       "      <th>Late_Aircraft</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>14</td>\n",
       "      <td>11</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>18</td>\n",
       "      <td>16</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>21</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>12</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>14</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "      <td>9</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>17</td>\n",
       "      <td>14</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>15</td>\n",
       "      <td>14</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>18</td>\n",
       "      <td>16</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Operating_Airline  Origin_Airport  Destination_Airport  \\\n",
       "0                  0               0                    8   \n",
       "1                  0               1                    0   \n",
       "2                  0               0                    1   \n",
       "3                  0               0                    5   \n",
       "4                  0               1                    2   \n",
       "5                  0               2                    1   \n",
       "6                  0               3                    1   \n",
       "7                  0               4                   17   \n",
       "8                  0               1                    5   \n",
       "9                  0               5                    1   \n",
       "\n",
       "   Scheduled_Arrival_Time  Scheduled_Departure_Time  Target  Month  Day  \\\n",
       "0                      14                        11       2      1    6   \n",
       "1                      18                        16       0      1    6   \n",
       "2                      21                        19       0      1    6   \n",
       "3                      12                        10       0      1    6   \n",
       "4                      11                        11       0      1    6   \n",
       "5                      14                        12       0      1    6   \n",
       "6                      10                         9       0      1    6   \n",
       "7                      14                        10       1      1    6   \n",
       "8                      15                        14       1      1    6   \n",
       "9                      18                        16       0      1    6   \n",
       "\n",
       "   Weekday  Carrier_Issue  Bad_Weather  NAS  Security_Issue  Late_Aircraft  \n",
       "0        3              1            0    0               0              0  \n",
       "1        3              0            0    0               0              0  \n",
       "2        3              0            0    0               0              0  \n",
       "3        3              0            0    0               0              0  \n",
       "4        3              0            0    0               0              0  \n",
       "5        3              0            0    0               0              0  \n",
       "6        3              0            0    0               0              0  \n",
       "7        3              0            0    0               0              0  \n",
       "8        3              0            0    0               0              0  \n",
       "9        3              0            0    0               0              0  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load DF as a Dask DF in order to use ML model on large dataset\n",
    "\n",
    "try:\n",
    "    if flights_df is not None:\n",
    "        flights_df = dd.from_pandas(flights_df, npartitions=2)\n",
    "except NameError:\n",
    "    model_df_path = Path('Resources/flights_model_df.csv')\n",
    "    flights_df = dd.read_csv(model_df_path)\n",
    "\n",
    "flights_df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1cca5c93",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Check datatypes\n",
    "\n",
    "flights_df.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "35aa606f",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\lady-\\anaconda3\\envs\\mlenv\\lib\\site-packages\\distributed\\node.py:182: UserWarning: Port 8787 is already in use.\n",
      "Perhaps you already have a cluster running?\n",
      "Hosting the HTTP server on port 52311 instead\n",
      "  warnings.warn(\n"
     ]
    }
   ],
   "source": [
    "# Start client for parallel computation\n",
    "\n",
    "client = Client()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1b6d88b1",
   "metadata": {},
   "source": [
    "## Random Forest Classifier Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f1c9552f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Function that runs a Random Forest Model and outputs stats on model\n",
    "\n",
    "def rfmodel_complete (MasterDF,ColumnList,estimators=100,randomstate=52):\n",
    "\n",
    "    X = MasterDF[ColumnList]\n",
    "    y = MasterDF['Target']\n",
    "\n",
    "    X_train, X_test, y_train, y_test = train_test_split(X,y,shuffle=True,random_state=randomstate)\n",
    "    \n",
    "    with parallel_backend('threading', n_jobs=2):\n",
    "        \n",
    "        rf_model = RandomForestClassifier(n_estimators=estimators,random_state=randomstate, n_jobs=2)\n",
    "        \n",
    "        rf_model.fit(X_train,y_train)\n",
    "            \n",
    "    rf_predictions = rf_model.predict(X_test)\n",
    "\n",
    "    cm = confusion_matrix(y_test, rf_predictions)\n",
    "\n",
    "    acc_score = accuracy_score(y_test, rf_predictions)\n",
    "\n",
    "    # Create a classification report\n",
    "    class_report = classification_report(y_test, rf_predictions)\n",
    "    \n",
    "    # Create a DataFrame from the confusion matrix.\n",
    "    cm_df = pd.DataFrame(\n",
    "        cm, index=[\"On Time\", \"Delayed\",\"Cancelled\"], columns=[\"On Time (Pred)\", \"Delayed (Pred)\", \"Cancelled (Pred)\"])\n",
    "\n",
    "    # Calculate feature importance in the Random Forest model.\n",
    "    importances = rf_model.feature_importances_\n",
    "\n",
    "    importances_dict = sorted(zip(importances, X.columns), reverse=True)\n",
    "    \n",
    "    return (# Displaying results\n",
    "        print(\"\\nConfusion Matrix\\n\"),\n",
    "        \n",
    "        print(cm_df),\n",
    "        \n",
    "        print(f\"\\nAccuracy Score : {acc_score}\\n\"),\n",
    "        \n",
    "        print(\"\\nClassification Report\\n\"),\n",
    "\n",
    "        print(class_report),\n",
    "        \n",
    "        \"Feature Importances\",\n",
    "        \n",
    "        importances_dict\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2e834de9",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "flights_df.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c191d11e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Run through with all features\n",
    "\n",
    "features = ['Operating_Airline', 'Origin_Airport', 'Destination_Airport',\n",
    "       'Scheduled_Arrival_Time', 'Scheduled_Departure_Time', 'Month',\n",
    "       'Day', 'Weekday', 'Carrier_Issue', 'Bad_Weather', 'NAS',\n",
    "       'Security_Issue', 'Late_Aircraft']\n",
    "\n",
    "rfmodel_complete(flights_df, features)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a0a7f2e2",
   "metadata": {},
   "source": [
    "## Balanced Random Classifier Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "f8a7dceb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Function that creates and fits a Balanced Random Forest Model and evaluates performance\n",
    "\n",
    "def balanced_rfmodel_complete (MasterDF,ColumnList,estimators=100,randomstate=52):\n",
    "\n",
    "    X = MasterDF[ColumnList]\n",
    "    y = MasterDF['Target']\n",
    "\n",
    "    X_train, X_test, y_train, y_test = train_test_split(X,y,shuffle=True,random_state=randomstate)\n",
    "    \n",
    "    with parallel_backend('threading', n_jobs=2):\n",
    "        \n",
    "        brf_model = BalancedRandomForestClassifier(n_estimators=estimators,random_state=randomstate, n_jobs=2)\n",
    "        \n",
    "        brf_model.fit(X_train,y_train)\n",
    "            \n",
    "    brf_predictions = brf_model.predict(X_test)\n",
    "\n",
    "    cm = confusion_matrix(y_test, brf_predictions)\n",
    "\n",
    "    acc_score = balanced_accuracy_score(y_test, brf_predictions)\n",
    "\n",
    "    # Create a classification report\n",
    "    class_report = classification_report_imbalanced(y_test, brf_predictions)\n",
    "    \n",
    "    # Create a DataFrame from the confusion matrix.\n",
    "    cm_df = pd.DataFrame(\n",
    "        cm, index=[\"On Time\", \"Delayed\",\"Cancelled\"], columns=[\"On Time (Pred)\", \"Delayed (Pred)\", \"Cancelled (Pred)\"])\n",
    "\n",
    "    # Calculate feature importance in the Random Forest model.\n",
    "    importances = brf_model.feature_importances_\n",
    "\n",
    "    importances_dict = sorted(zip(importances, X.columns), reverse=True)\n",
    "    \n",
    "    return (# Displaying results\n",
    "        print(\"\\nConfusion Matrix\\n\"),\n",
    "        \n",
    "        print(cm_df),\n",
    "        \n",
    "        print(f\"\\nAccuracy Score : {acc_score}\\n\"),\n",
    "        \n",
    "        print(\"\\nClassification Report\\n\"),\n",
    "\n",
    "        print(class_report),\n",
    "        \n",
    "        \"Feature Importances\",\n",
    "        \n",
    "        importances_dict\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "56dbbc5e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Confusion Matrix\n",
      "\n",
      "           On Time (Pred)  Delayed (Pred)  Cancelled (Pred)\n",
      "On Time            183675           21206                 0\n",
      "Delayed             83268           89341             19339\n",
      "Cancelled               0             473             11484\n",
      "\n",
      "Accuracy Score : 0.7741271210388193\n",
      "\n",
      "\n",
      "Classification Report\n",
      "\n",
      "                   pre       rec       spe        f1       geo       iba       sup\n",
      "\n",
      "          0       0.69      0.90      0.59      0.78      0.73      0.55    204881\n",
      "          1       0.80      0.47      0.90      0.59      0.65      0.40    191948\n",
      "          2       0.37      0.96      0.95      0.54      0.96      0.91     11957\n",
      "\n",
      "avg / total       0.73      0.70      0.75      0.68      0.70      0.49    408786\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " 'Feature Importances',\n",
       " [(0.21440226183321123, 'Bad_Weather'),\n",
       "  (0.12421643154957636, 'Carrier_Issue'),\n",
       "  (0.10857299924064082, 'Destination_Airport'),\n",
       "  (0.10724793210459607, 'Origin_Airport'),\n",
       "  (0.09154050593191856, 'Day'),\n",
       "  (0.0684719823909734, 'Operating_Airline'),\n",
       "  (0.0668296349347912, 'NAS'),\n",
       "  (0.058691194045272635, 'Month'),\n",
       "  (0.05834953585328824, 'Scheduled_Departure_Time'),\n",
       "  (0.055487429294683734, 'Scheduled_Arrival_Time'),\n",
       "  (0.04165998499991177, 'Weekday'),\n",
       "  (0.0025137505812640737, 'Late_Aircraft'),\n",
       "  (0.0020163572398719117, 'Security_Issue')])"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Run through with all features\n",
    "\n",
    "features = ['Operating_Airline', 'Origin_Airport', 'Destination_Airport',\n",
    "       'Scheduled_Arrival_Time', 'Scheduled_Departure_Time', 'Month',\n",
    "       'Day', 'Weekday', 'Carrier_Issue', 'Bad_Weather', 'NAS',\n",
    "       'Security_Issue', 'Late_Aircraft']\n",
    "\n",
    "balanced_rfmodel_complete(flights_df, features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "98df72a9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Confusion Matrix\n",
      "\n",
      "           On Time (Pred)  Delayed (Pred)  Cancelled (Pred)\n",
      "On Time            182518           22068               295\n",
      "Delayed             81347           91033             19568\n",
      "Cancelled              28             478             11451\n",
      "\n",
      "Accuracy Score : 0.7742630601062999\n",
      "\n",
      "\n",
      "Classification Report\n",
      "\n",
      "                   pre       rec       spe        f1       geo       iba       sup\n",
      "\n",
      "          0       0.69      0.89      0.60      0.78      0.73      0.55    204881\n",
      "          1       0.80      0.47      0.90      0.60      0.65      0.41    191948\n",
      "          2       0.37      0.96      0.95      0.53      0.95      0.91     11957\n",
      "\n",
      "avg / total       0.73      0.70      0.75      0.69      0.70      0.49    408786\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " 'Feature Importances',\n",
       " [(0.2142600982444385, 'Bad_Weather'),\n",
       "  (0.13699789472881715, 'Day'),\n",
       "  (0.12928672203148353, 'Carrier_Issue'),\n",
       "  (0.1074853263230668, 'Destination_Airport'),\n",
       "  (0.10526817750008494, 'Origin_Airport'),\n",
       "  (0.0703253836179386, 'NAS'),\n",
       "  (0.06702502078302643, 'Month'),\n",
       "  (0.06215193234928705, 'Operating_Airline'),\n",
       "  (0.055505445101277064, 'Scheduled_Departure_Time'),\n",
       "  (0.05169399932057998, 'Scheduled_Arrival_Time')])"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Run through with features < 0.05 removed\n",
    "\n",
    "features = ['Operating_Airline', 'Origin_Airport', 'Destination_Airport',\n",
    "       'Scheduled_Arrival_Time', 'Scheduled_Departure_Time', 'Month',\n",
    "       'Day', 'Carrier_Issue', 'Bad_Weather', 'NAS']\n",
    "\n",
    "balanced_rfmodel_complete(flights_df, features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "af39c175",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RFC Model for 16 number of estimators:\n",
      "\n",
      "Confusion Matrix\n",
      "\n",
      "           On Time (Pred)  Delayed (Pred)  Cancelled (Pred)\n",
      "On Time            176917           27685               279\n",
      "Delayed             81742           91653             18553\n",
      "Cancelled              31             656             11270\n",
      "\n",
      "Accuracy Score : 0.7611812749941386\n",
      "\n",
      "\n",
      "Classification Report\n",
      "\n",
      "                   pre       rec       spe        f1       geo       iba       sup\n",
      "\n",
      "          0       0.68      0.86      0.60      0.76      0.72      0.53    204881\n",
      "          1       0.76      0.48      0.87      0.59      0.64      0.40    191948\n",
      "          2       0.37      0.94      0.95      0.54      0.95      0.90     11957\n",
      "\n",
      "avg / total       0.71      0.68      0.74      0.67      0.69      0.48    408786\n",
      "\n",
      "================================================\n",
      "RFC Model for 32 number of estimators:\n",
      "\n",
      "Confusion Matrix\n",
      "\n",
      "           On Time (Pred)  Delayed (Pred)  Cancelled (Pred)\n",
      "On Time            179555           25046               280\n",
      "Delayed             81225           91642             19081\n",
      "Cancelled              30             540             11387\n",
      "\n",
      "Accuracy Score : 0.7687157826155256\n",
      "\n",
      "\n",
      "Classification Report\n",
      "\n",
      "                   pre       rec       spe        f1       geo       iba       sup\n",
      "\n",
      "          0       0.69      0.88      0.60      0.77      0.73      0.54    204881\n",
      "          1       0.78      0.48      0.88      0.59      0.65      0.40    191948\n",
      "          2       0.37      0.95      0.95      0.53      0.95      0.91     11957\n",
      "\n",
      "avg / total       0.72      0.69      0.74      0.68      0.70      0.49    408786\n",
      "\n",
      "================================================\n",
      "RFC Model for 64 number of estimators:\n",
      "\n",
      "Confusion Matrix\n",
      "\n",
      "           On Time (Pred)  Delayed (Pred)  Cancelled (Pred)\n",
      "On Time            181643           22958               280\n",
      "Delayed             81243           91277             19428\n",
      "Cancelled              28             496             11433\n",
      "\n",
      "Accuracy Score : 0.7727613972619797\n",
      "\n",
      "\n",
      "Classification Report\n",
      "\n",
      "                   pre       rec       spe        f1       geo       iba       sup\n",
      "\n",
      "          0       0.69      0.89      0.60      0.78      0.73      0.55    204881\n",
      "          1       0.80      0.48      0.89      0.60      0.65      0.41    191948\n",
      "          2       0.37      0.96      0.95      0.53      0.95      0.91     11957\n",
      "\n",
      "avg / total       0.73      0.70      0.75      0.68      0.70      0.49    408786\n",
      "\n",
      "================================================\n",
      "RFC Model for 100 number of estimators:\n",
      "\n",
      "Confusion Matrix\n",
      "\n",
      "           On Time (Pred)  Delayed (Pred)  Cancelled (Pred)\n",
      "On Time            182518           22068               295\n",
      "Delayed             81347           91033             19568\n",
      "Cancelled              28             478             11451\n",
      "\n",
      "Accuracy Score : 0.7742630601062999\n",
      "\n",
      "\n",
      "Classification Report\n",
      "\n",
      "                   pre       rec       spe        f1       geo       iba       sup\n",
      "\n",
      "          0       0.69      0.89      0.60      0.78      0.73      0.55    204881\n",
      "          1       0.80      0.47      0.90      0.60      0.65      0.41    191948\n",
      "          2       0.37      0.96      0.95      0.53      0.95      0.91     11957\n",
      "\n",
      "avg / total       0.73      0.70      0.75      0.69      0.70      0.49    408786\n",
      "\n",
      "================================================\n",
      "RFC Model for 200 number of estimators:\n",
      "\n",
      "Confusion Matrix\n",
      "\n",
      "           On Time (Pred)  Delayed (Pred)  Cancelled (Pred)\n",
      "On Time            183275           21305               301\n",
      "Delayed             81456           90942             19550\n",
      "Cancelled              28             460             11469\n",
      "\n",
      "Accuracy Score : 0.775838438548675\n",
      "\n",
      "\n",
      "Classification Report\n",
      "\n",
      "                   pre       rec       spe        f1       geo       iba       sup\n",
      "\n",
      "          0       0.69      0.89      0.60      0.78      0.73      0.55    204881\n",
      "          1       0.81      0.47      0.90      0.60      0.65      0.41    191948\n",
      "          2       0.37      0.96      0.95      0.53      0.95      0.91     11957\n",
      "\n",
      "avg / total       0.74      0.70      0.75      0.69      0.70      0.50    408786\n",
      "\n",
      "================================================\n"
     ]
    }
   ],
   "source": [
    "# Test different numbers of estimators\n",
    "n_estimators = [16, 32, 64, 100, 200]\n",
    "\n",
    "features = ['Operating_Airline', 'Origin_Airport', 'Destination_Airport',\n",
    "       'Scheduled_Arrival_Time', 'Scheduled_Departure_Time', 'Month',\n",
    "       'Day', 'Carrier_Issue', 'Bad_Weather', 'NAS']\n",
    "for estimator in n_estimators:\n",
    "    \n",
    "    print(f'RFC Model for {estimator} number of estimators:')\n",
    "    balanced_rfmodel_complete(flights_df, features, estimators=estimator)\n",
    "    print('================================================')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "21fba1eb",
   "metadata": {},
   "source": [
    "## Visualize Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "eaee1b0a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "a1a9f824",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Rerun model with best parameters and get confusion matrix\n",
    "\n",
    "X = flights_df[['Operating_Airline', 'Origin_Airport', 'Destination_Airport',\n",
    "       'Scheduled_Arrival_Time', 'Scheduled_Departure_Time', 'Month',\n",
    "       'Day', 'Carrier_Issue', 'Bad_Weather', 'NAS']]\n",
    "y = flights_df['Target']\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, shuffle=True, random_state=52)\n",
    "\n",
    "with parallel_backend('threading', n_jobs=2):\n",
    "\n",
    "    brf_model = BalancedRandomForestClassifier(n_estimators=64, random_state=52, n_jobs=2)\n",
    "\n",
    "    brf_model.fit(X_train, y_train)\n",
    "\n",
    "brf_predictions = brf_model.predict(X_test)\n",
    "cm = confusion_matrix(y_test, brf_predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "516c1a9e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Make labels for the heatmap\n",
    "\n",
    "counts = [f'{val}' for val in cm.flatten()]\n",
    "percents = [f'{val:.2%}' for val in cm.flatten()/np.sum(cm)]\n",
    "labels = [f'{count}\\n{percent}' for count, percent in zip(counts,percents)]\n",
    "\n",
    "# Generate the heatmap with labels as 3x3 array\n",
    "\n",
    "cm_heatmap = sns.heatmap(cm/np.sum(cm), annot=np.asarray(labels).reshape(3, 3),\n",
    "                         fmt='', cmap='Blues',\n",
    "                         xticklabels=['On-time (Pred)', 'Delayed (Pred)', 'Cancelled (Pred)'],\n",
    "                         yticklabels=['On-time', 'Delayed', 'Cancelled'])\n",
    "\n",
    "# Save heatmap to Resources/images\n",
    "heatmap_path = Path('Resources/images/confusion_matrix_heatmap.png')\n",
    "cm_heatmap.get_figure().savefig(heatmap_path)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "mlenv",
   "language": "python",
   "name": "mlenv"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}