NonNumericFeaturelinear.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Linear regression single non-numeric feature \n",
    "## One-Hot encoding\n",
    "\n",
    "This notebook contains code to train and evaluate a linear regression model using a single non-numeric feature and label.\n",
    "\n",
    "The technique we will use to handle the non-numeric features is called [One-Hot encoding](https://en.wikipedia.org/wiki/One-hot)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Import the required python libraries\n",
    "- **pandas** contains the dataframe object and a number of useful methods for manipulating and querying data contained in a dataframe\n",
    "- **numpy** contains a number of useful mathematical operations including some that are helpful when evaluating accuracy of trained models\n",
    "- **Scikitlearn train_test_split** splits data into training and test sets\n",
    "- **scikitlearn LinearRegression** used to train a linear regression model\n",
    "- **scikitlearn metrics** used to calculate metrics such as Mean Squared Error, helpful when evaluating accuracy of trained models\n",
    "- **matplotlib pyplot** used to plot graphs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn import metrics\n",
    "#import statsmodels.api as sm\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Read flight data from csv file into Pandas dataframe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(616101, 17)"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "flight_df=pd.read_csv('all_flights.csv') # Read csv file into flight_df dataframe\n",
    "flight_df.shape                          # Display shape of array to see how many rows and columns are in the dataframe"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Display the first 10 rows in the dataset to make sure data looks like it imported correctly"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>FL_DATE</th>\n",
       "      <th>OP_UNIQUE_CARRIER</th>\n",
       "      <th>TAIL_NUM</th>\n",
       "      <th>OP_CARRIER_FL_NUM</th>\n",
       "      <th>ORIGIN</th>\n",
       "      <th>DEST</th>\n",
       "      <th>CRS_DEP_TIME</th>\n",
       "      <th>DEP_TIME</th>\n",
       "      <th>DEP_DELAY</th>\n",
       "      <th>CRS_ARR_TIME</th>\n",
       "      <th>ARR_TIME</th>\n",
       "      <th>ARR_DELAY</th>\n",
       "      <th>CRS_ELAPSED_TIME</th>\n",
       "      <th>ACTUAL_ELAPSED_TIME</th>\n",
       "      <th>AIR_TIME</th>\n",
       "      <th>DISTANCE</th>\n",
       "      <th>Unnamed: 16</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2018-10-01</td>\n",
       "      <td>WN</td>\n",
       "      <td>N221WN</td>\n",
       "      <td>802</td>\n",
       "      <td>ABQ</td>\n",
       "      <td>BWI</td>\n",
       "      <td>905</td>\n",
       "      <td>903.0</td>\n",
       "      <td>-2.0</td>\n",
       "      <td>1450</td>\n",
       "      <td>1433.0</td>\n",
       "      <td>-17.0</td>\n",
       "      <td>225.0</td>\n",
       "      <td>210.0</td>\n",
       "      <td>197.0</td>\n",
       "      <td>1670.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2018-10-01</td>\n",
       "      <td>WN</td>\n",
       "      <td>N8329B</td>\n",
       "      <td>3744</td>\n",
       "      <td>ABQ</td>\n",
       "      <td>BWI</td>\n",
       "      <td>1500</td>\n",
       "      <td>1458.0</td>\n",
       "      <td>-2.0</td>\n",
       "      <td>2045</td>\n",
       "      <td>2020.0</td>\n",
       "      <td>-25.0</td>\n",
       "      <td>225.0</td>\n",
       "      <td>202.0</td>\n",
       "      <td>191.0</td>\n",
       "      <td>1670.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2018-10-01</td>\n",
       "      <td>WN</td>\n",
       "      <td>N920WN</td>\n",
       "      <td>1019</td>\n",
       "      <td>ABQ</td>\n",
       "      <td>DAL</td>\n",
       "      <td>1800</td>\n",
       "      <td>1802.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2045</td>\n",
       "      <td>2032.0</td>\n",
       "      <td>-13.0</td>\n",
       "      <td>105.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>80.0</td>\n",
       "      <td>580.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2018-10-01</td>\n",
       "      <td>WN</td>\n",
       "      <td>N480WN</td>\n",
       "      <td>1499</td>\n",
       "      <td>ABQ</td>\n",
       "      <td>DAL</td>\n",
       "      <td>950</td>\n",
       "      <td>947.0</td>\n",
       "      <td>-3.0</td>\n",
       "      <td>1235</td>\n",
       "      <td>1223.0</td>\n",
       "      <td>-12.0</td>\n",
       "      <td>105.0</td>\n",
       "      <td>96.0</td>\n",
       "      <td>81.0</td>\n",
       "      <td>580.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2018-10-01</td>\n",
       "      <td>WN</td>\n",
       "      <td>N227WN</td>\n",
       "      <td>3635</td>\n",
       "      <td>ABQ</td>\n",
       "      <td>DAL</td>\n",
       "      <td>1150</td>\n",
       "      <td>1151.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1430</td>\n",
       "      <td>1423.0</td>\n",
       "      <td>-7.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>92.0</td>\n",
       "      <td>80.0</td>\n",
       "      <td>580.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      FL_DATE OP_UNIQUE_CARRIER TAIL_NUM  OP_CARRIER_FL_NUM ORIGIN DEST  \\\n",
       "0  2018-10-01                WN   N221WN                802    ABQ  BWI   \n",
       "1  2018-10-01                WN   N8329B               3744    ABQ  BWI   \n",
       "2  2018-10-01                WN   N920WN               1019    ABQ  DAL   \n",
       "3  2018-10-01                WN   N480WN               1499    ABQ  DAL   \n",
       "4  2018-10-01                WN   N227WN               3635    ABQ  DAL   \n",
       "\n",
       "   CRS_DEP_TIME  DEP_TIME  DEP_DELAY  CRS_ARR_TIME  ARR_TIME  ARR_DELAY  \\\n",
       "0           905     903.0       -2.0          1450    1433.0      -17.0   \n",
       "1          1500    1458.0       -2.0          2045    2020.0      -25.0   \n",
       "2          1800    1802.0        2.0          2045    2032.0      -13.0   \n",
       "3           950     947.0       -3.0          1235    1223.0      -12.0   \n",
       "4          1150    1151.0        1.0          1430    1423.0       -7.0   \n",
       "\n",
       "   CRS_ELAPSED_TIME  ACTUAL_ELAPSED_TIME  AIR_TIME  DISTANCE  Unnamed: 16  \n",
       "0             225.0                210.0     197.0    1670.0          NaN  \n",
       "1             225.0                202.0     191.0    1670.0          NaN  \n",
       "2             105.0                 90.0      80.0     580.0          NaN  \n",
       "3             105.0                 96.0      81.0     580.0          NaN  \n",
       "4             100.0                 92.0      80.0     580.0          NaN  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "flight_df.head()                         # Displays the top 10 rows from flight_df dataframe"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Display the names of the columns in the dataframe and their datatypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "FL_DATE                 object\n",
       "OP_UNIQUE_CARRIER       object\n",
       "TAIL_NUM                object\n",
       "OP_CARRIER_FL_NUM        int64\n",
       "ORIGIN                  object\n",
       "DEST                    object\n",
       "CRS_DEP_TIME             int64\n",
       "DEP_TIME               float64\n",
       "DEP_DELAY              float64\n",
       "CRS_ARR_TIME             int64\n",
       "ARR_TIME               float64\n",
       "ARR_DELAY              float64\n",
       "CRS_ELAPSED_TIME       float64\n",
       "ACTUAL_ELAPSED_TIME    float64\n",
       "AIR_TIME               float64\n",
       "DISTANCE               float64\n",
       "Unnamed: 16            float64\n",
       "dtype: object"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "flight_df.dtypes  # Displays the names and data types of the columns in the dataframe"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create a new dataframe containing ONLY the columns we want to use as features and labels.\n",
    "This saves us spending time cleaning up data we are not using to train our model.\n",
    "This model will use OP_UNIQUE_CARRIER (the airline operating the flight) as a feature to predict the value of the label ARR_DELAY"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(616101, 2)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "min_flight_data_df = flight_df[['OP_UNIQUE_CARRIER','ARR_DELAY']] # Create a new dataframe containing only\n",
    "                                                                  # OP_UNIQUE_CARRIER and ARR_DELAY\n",
    "min_flight_data_df.shape                                          # Display the shape of the dataframe as a quick check to \n",
    "                                                                  # ensure dataframe has expected number of columns and rows"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Get rid of rows containing NaN/missing values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(610334, 2)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "no_missing_values_df = min_flight_data_df.dropna(axis=0,how='any')  # Use dropna to remove rows with NaN in any column\n",
    "no_missing_values_df.shape                                          # Display shape to see how many rows were removed"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a closer look at our data and find out how many flights we have from each carrier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OP_UNIQUE_CARRIER\n",
       "9E     20819\n",
       "AA     78092\n",
       "AS     21396\n",
       "B6     24716\n",
       "DL     81950\n",
       "EV     14993\n",
       "F9     10320\n",
       "G4      7559\n",
       "HA      7137\n",
       "MQ     25548\n",
       "NK     13725\n",
       "OH     23459\n",
       "OO     66063\n",
       "UA     55388\n",
       "WN    113840\n",
       "YV     18728\n",
       "YX     26601\n",
       "Name: ARR_DELAY, dtype: int64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "no_missing_values_df.groupby('OP_UNIQUE_CARRIER')['ARR_DELAY'].count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can also use a graph to display the number of flights by each carrier visually"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Nbr Flights')"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = no_missing_values_df.groupby('OP_UNIQUE_CARRIER')['ARR_DELAY'].count().plot(kind='bar', title='Flights by Carrier')\n",
    "ax.set_xlabel('Carrier')\n",
    "ax.set_ylabel('Nbr Flights')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Continuing to explore our data, let's check the average delay time by carrier to see if there is any correlation between carriers and arrival delay"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OP_UNIQUE_CARRIER\n",
       "9E     1.074403\n",
       "AA     6.403908\n",
       "AS     0.992008\n",
       "B6     4.337797\n",
       "DL    -3.546894\n",
       "EV    10.948643\n",
       "F9    13.964922\n",
       "G4     4.158883\n",
       "HA    -0.595208\n",
       "MQ     2.816541\n",
       "NK    -2.920583\n",
       "OH     0.657530\n",
       "OO     3.772550\n",
       "UA     3.397740\n",
       "WN     1.927837\n",
       "YV    14.740709\n",
       "YX     1.803992\n",
       "Name: ARR_DELAY, dtype: float64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "no_missing_values_df.groupby('OP_UNIQUE_CARRIER')['ARR_DELAY'].mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's display that same data visually in a graph"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Average Flight Delay')"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax=no_missing_values_df.groupby('OP_UNIQUE_CARRIER')['ARR_DELAY'].mean().plot(kind='bar',title='Average delay by carrier')\n",
    "ax.set_xlabel('Carrier')\n",
    "ax.set_ylabel('Average Flight Delay')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There does seem to be a relation between carrier and delay times. So maybe I can train a model using OP_UNIQUE_CARRIER as a feature.\n",
    "\n",
    "OP_UNIQUE_CARRIER is not numeric so I need to make it numeric, basically I want to know iif a flight is operated by a particular airline is it more likely to be late. \n",
    "\n",
    "I need to make a dummy column for each possible airline i.e. AIRCANADA 1/0 WESTJET 1/0 AIRTRANSAT 1/0 for every row. a value of 1 means the flight was operated by that carrier. Then I can include those columns as features when I train my model.\n",
    "\n",
    "The *[get_dummies](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.get_dummies.html)* method in pandas will return a dataframe of dummy columns for a given column.\n",
    "\n",
    "This particular technique for converting string to numeric data by creating dummy columns for each possible value and specifying a 1 or 0 to indicate which rows contain which value is called one-hot encoding."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<bound method NDFrame.head of         9E  AA  AS  B6  DL  EV  F9  G4  HA  MQ  NK  OH  OO  UA  WN  YV  YX\n",
       "0        0   0   0   0   0   0   0   0   0   0   0   0   0   0   1   0   0\n",
       "1        0   0   0   0   0   0   0   0   0   0   0   0   0   0   1   0   0\n",
       "2        0   0   0   0   0   0   0   0   0   0   0   0   0   0   1   0   0\n",
       "3        0   0   0   0   0   0   0   0   0   0   0   0   0   0   1   0   0\n",
       "4        0   0   0   0   0   0   0   0   0   0   0   0   0   0   1   0   0\n",
       "...     ..  ..  ..  ..  ..  ..  ..  ..  ..  ..  ..  ..  ..  ..  ..  ..  ..\n",
       "616096   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   1   0\n",
       "616097   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   1   0\n",
       "616098   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   1   0\n",
       "616099   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   1   0\n",
       "616100   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   1   0\n",
       "\n",
       "[610334 rows x 17 columns]>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dummy_df = pd.get_dummies(no_missing_values_df['OP_UNIQUE_CARRIER'])\n",
    "dummy_df.head"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create one dataframe containing only the features OP_UNIQUE_CARRIER dummy columns, and one dataframe containing only the label (ARR_DELAY) \n",
    "If either of these has only one column you must reshape the dataframe to -1,1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a dataFrame containing the feature columns\n",
    "X = dummy_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a DataFrame containing the labels\n",
    "# Reshape to -1,1 if only containing a single column\n",
    "y = no_missing_values_df['ARR_DELAY'].values.reshape(-1,1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Split the data into two datasets, one for training the model, one for testing the model\n",
    "\n",
    "scikitlearn method [train_test_split]( \n",
    "https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.train_test_split.html) \n",
    "- **X** is the dataframe containing our features\n",
    "- **Y** is the dataframe containing our label\n",
    "- **test_size** determines what fraction of the data is put into the test dataframe\n",
    "- **random_state** defaults to a random number, by specifying a specific number I ensure the split I generate is reproducible. That way if I make changes I know changes in accuracy are not caused by different rows used as training or test data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Split into training and test data\n",
    "X_train, X_test, y_train, y_test = train_test_split(X,y,test_size=0.30, random_state=42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Check the size of your test and training datasets (i.e. how many rows in each)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(427233, 17)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(183101, 17)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train the model using scikitlearn [LinearRegression]( \n",
    "https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html)\n",
    "\n",
    "Your feature must be numeric that's why we created the dummy columns for OP_UNIQUE_CARRIER\n",
    "\n",
    "Your training data cannot contain any missing values in rows."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearRegression(copy_X=True, fit_intercept=True, n_jobs=None, normalize=False)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "regressor = LinearRegression()     # Create a scikit learn LinearRegression object\n",
    "regressor.fit(X_train, y_train)    # Use the fit methong to train the model using your training data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's see what arrival delays are predicted for our test data by using the **predict** method of our linearRegression object"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred = regressor.predict(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's compare the actual arrival delay times and the predicted arrival delay times"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<bound method NDFrame.head of         Actual  Predicted\n",
       "0        -13.0  -3.507225\n",
       "1        -24.0  14.829079\n",
       "2        100.0   6.300830\n",
       "3         -8.0   0.796358\n",
       "4        -19.0  13.776355\n",
       "...        ...        ...\n",
       "183096    48.0   1.357497\n",
       "183097   -24.0   1.921001\n",
       "183098    20.0   1.939634\n",
       "183099    47.0  14.829079\n",
       "183100   -11.0   1.939634\n",
       "\n",
       "[183101 rows x 2 columns]>"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#combine the two 1D numpy arrays into a 2D array\n",
    "combined = np.hstack((y_test,y_pred))\n",
    "#Convert to a DataFrame\n",
    "accuracy_df = pd.DataFrame(combined,columns=['Actual','Predicted'])\n",
    "accuracy_df.head"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can do some calculations to get a sense of overall accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mean absolute error:  20.10786357633716\n"
     ]
    }
   ],
   "source": [
    "print('Mean absolute error: ',metrics.mean_absolute_error(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "print('Mean Squared Error:', metrics.mean_squared_error(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Root Mean Squared Error: 41.91637637662109\n"
     ]
    }
   ],
   "source": [
    "print('Root Mean Squared Error:', np.sqrt(metrics.mean_squared_error(y_test,y_pred)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Check coefficients for our attributes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[-7.66451618e+09 -7.66451618e+09 -7.66451618e+09 -7.66451618e+09\n",
      "  -7.66451619e+09 -7.66451617e+09 -7.66451617e+09 -7.66451618e+09\n",
      "  -7.66451619e+09 -7.66451618e+09 -7.66451619e+09 -7.66451618e+09\n",
      "  -7.66451618e+09 -7.66451618e+09 -7.66451618e+09 -7.66451617e+09\n",
      "  -7.66451618e+09]]\n",
      "Index(['9E', 'AA', 'AS', 'B6', 'DL', 'EV', 'F9', 'G4', 'HA', 'MQ', 'NK', 'OH',\n",
      "       'OO', 'UA', 'WN', 'YV', 'YX'],\n",
      "      dtype='object')\n"
     ]
    }
   ],
   "source": [
    "print(regressor.coef_)\n",
    "print(X_train.columns)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "These are **very** small numbers so it does not appear there is a strong correlation between airline and number of minutes late for the flight."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}