Predicting House Prices with Keras.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## House Price Prediction Using Keras\n",
    "\n",
    "---\n",
    "\n",
    "For this project, I am going to predict price of houses given the following features:\n",
    "\n",
    "1. Year of sale of the house\n",
    "2. The age of the house at the time of sale\n",
    "3. Distance from city center\n",
    "4. Number of stores in the locality\n",
    "5. The latitude\n",
    "6. The longitude\n",
    "\n",
    "![Regression](Images/regression.jpeg)\n",
    "\n",
    "Note: This notebook uses `python 3` and these packages: `tensorflow`, `pandas`, `matplotlib`, `scikit-learn`."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importing Libraries & Helper Functions\n",
    "\n",
    "First of all, we will need to import some libraries and helper functions. This includes TensorFlow and some utility functions that I've written to save time."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Libraries imported.\n"
     ]
    }
   ],
   "source": [
    "#Importing the relevant libraries\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import tensorflow as tf\n",
    "\n",
    "from utils import *\n",
    "from sklearn.model_selection import train_test_split\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Dropout\n",
    "from tensorflow.keras.callbacks import EarlyStopping, LambdaCallback\n",
    "\n",
    "%matplotlib inline\n",
    "tf.logging.set_verbosity(tf.logging.ERROR)\n",
    "\n",
    "print('Libraries imported.')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importing the Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.1: Importing the Data\n",
    "\n",
    "The dataset is saved in a `data.csv` file. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>serial</th>\n",
       "      <th>date</th>\n",
       "      <th>age</th>\n",
       "      <th>distance</th>\n",
       "      <th>stores</th>\n",
       "      <th>latitude</th>\n",
       "      <th>longitude</th>\n",
       "      <th>price</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>2009</td>\n",
       "      <td>21</td>\n",
       "      <td>9</td>\n",
       "      <td>6</td>\n",
       "      <td>84</td>\n",
       "      <td>121</td>\n",
       "      <td>14264</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>2007</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>86</td>\n",
       "      <td>121</td>\n",
       "      <td>12032</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2016</td>\n",
       "      <td>18</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>90</td>\n",
       "      <td>120</td>\n",
       "      <td>13560</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>2002</td>\n",
       "      <td>13</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>80</td>\n",
       "      <td>128</td>\n",
       "      <td>12029</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>2014</td>\n",
       "      <td>25</td>\n",
       "      <td>5</td>\n",
       "      <td>8</td>\n",
       "      <td>81</td>\n",
       "      <td>122</td>\n",
       "      <td>14157</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   serial  date  age  distance  stores  latitude  longitude  price\n",
       "0       0  2009   21         9       6        84        121  14264\n",
       "1       1  2007    4         2       3        86        121  12032\n",
       "2       2  2016   18         3       7        90        120  13560\n",
       "3       3  2002   13         2       2        80        128  12029\n",
       "4       4  2014   25         5       8        81        122  14157"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv(\"data.csv\",names = column_names)\n",
    "df.head()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.2: Check Missing Data\n",
    "\n",
    "Now, let us check if the data has any missing values. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "serial       0\n",
       "date         0\n",
       "age          0\n",
       "distance     0\n",
       "stores       0\n",
       "latitude     0\n",
       "longitude    0\n",
       "price        0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#len(df) # 5000 rows\n",
    "\n",
    "df.isna().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Normalization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.1: Data Normalization\n",
    "\n",
    "Here I am making it easier for optimization algorithms to find minimas by normalizing the data before training a model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>date</th>\n",
       "      <th>age</th>\n",
       "      <th>distance</th>\n",
       "      <th>stores</th>\n",
       "      <th>latitude</th>\n",
       "      <th>longitude</th>\n",
       "      <th>price</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.015978</td>\n",
       "      <td>0.181384</td>\n",
       "      <td>1.257002</td>\n",
       "      <td>0.345224</td>\n",
       "      <td>-0.307212</td>\n",
       "      <td>-1.260799</td>\n",
       "      <td>0.350088</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.350485</td>\n",
       "      <td>-1.319118</td>\n",
       "      <td>-0.930610</td>\n",
       "      <td>-0.609312</td>\n",
       "      <td>0.325301</td>\n",
       "      <td>-1.260799</td>\n",
       "      <td>-1.836486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.298598</td>\n",
       "      <td>-0.083410</td>\n",
       "      <td>-0.618094</td>\n",
       "      <td>0.663402</td>\n",
       "      <td>1.590328</td>\n",
       "      <td>-1.576456</td>\n",
       "      <td>-0.339584</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-1.266643</td>\n",
       "      <td>-0.524735</td>\n",
       "      <td>-0.930610</td>\n",
       "      <td>-0.927491</td>\n",
       "      <td>-1.572238</td>\n",
       "      <td>0.948803</td>\n",
       "      <td>-1.839425</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.932135</td>\n",
       "      <td>0.534444</td>\n",
       "      <td>0.006938</td>\n",
       "      <td>0.981581</td>\n",
       "      <td>-1.255981</td>\n",
       "      <td>-0.945141</td>\n",
       "      <td>0.245266</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       date       age  distance    stores  latitude  longitude     price\n",
       "0  0.015978  0.181384  1.257002  0.345224 -0.307212  -1.260799  0.350088\n",
       "1 -0.350485 -1.319118 -0.930610 -0.609312  0.325301  -1.260799 -1.836486\n",
       "2  1.298598 -0.083410 -0.618094  0.663402  1.590328  -1.576456 -0.339584\n",
       "3 -1.266643 -0.524735 -0.930610 -0.927491 -1.572238   0.948803 -1.839425\n",
       "4  0.932135  0.534444  0.006938  0.981581 -1.255981  -0.945141  0.245266"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = df.iloc[:, 1:] # Ignoring the \"Serial\" column\n",
    "df_norm = (df - df.mean())/df.std()\n",
    "df_norm.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.2: Convert Label Value\n",
    "\n",
    "Because we are using normalized values for the labels, we will get the predictions back from a trained model in the same distribution. So, we need to convert the predicted values back to the original distribution if we want predicted prices."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "14263\n"
     ]
    }
   ],
   "source": [
    "y_mean = df['price'].mean()\n",
    "y_std = df['price'].std()\n",
    "\n",
    "def convert_label_value(pred):\n",
    "    return int(pred*y_std + y_mean)\n",
    "\n",
    "print(convert_label_value(0.350088))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This price, $14,263 is the same (approximately) as the original price value."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create Training and Test Sets"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.1: Select Features\n",
    "\n",
    "Now, we will remove the column __price__ from the list of features as it is the label and should not be used as a feature."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>date</th>\n",
       "      <th>age</th>\n",
       "      <th>distance</th>\n",
       "      <th>stores</th>\n",
       "      <th>latitude</th>\n",
       "      <th>longitude</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.015978</td>\n",
       "      <td>0.181384</td>\n",
       "      <td>1.257002</td>\n",
       "      <td>0.345224</td>\n",
       "      <td>-0.307212</td>\n",
       "      <td>-1.260799</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.350485</td>\n",
       "      <td>-1.319118</td>\n",
       "      <td>-0.930610</td>\n",
       "      <td>-0.609312</td>\n",
       "      <td>0.325301</td>\n",
       "      <td>-1.260799</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.298598</td>\n",
       "      <td>-0.083410</td>\n",
       "      <td>-0.618094</td>\n",
       "      <td>0.663402</td>\n",
       "      <td>1.590328</td>\n",
       "      <td>-1.576456</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-1.266643</td>\n",
       "      <td>-0.524735</td>\n",
       "      <td>-0.930610</td>\n",
       "      <td>-0.927491</td>\n",
       "      <td>-1.572238</td>\n",
       "      <td>0.948803</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.932135</td>\n",
       "      <td>0.534444</td>\n",
       "      <td>0.006938</td>\n",
       "      <td>0.981581</td>\n",
       "      <td>-1.255981</td>\n",
       "      <td>-0.945141</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       date       age  distance    stores  latitude  longitude\n",
       "0  0.015978  0.181384  1.257002  0.345224 -0.307212  -1.260799\n",
       "1 -0.350485 -1.319118 -0.930610 -0.609312  0.325301  -1.260799\n",
       "2  1.298598 -0.083410 -0.618094  0.663402  1.590328  -1.576456\n",
       "3 -1.266643 -0.524735 -0.930610 -0.927491 -1.572238   0.948803\n",
       "4  0.932135  0.534444  0.006938  0.981581 -1.255981  -0.945141"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = df_norm.iloc[:,:6]\n",
    "X.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.2: Select Labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    0.350088\n",
       "1   -1.836486\n",
       "2   -0.339584\n",
       "3   -1.839425\n",
       "4    0.245266\n",
       "Name: price, dtype: float64"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y = df_norm.iloc[:,-1]\n",
    "y.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.3: Feature and Label Values\n",
    "\n",
    "We will need to extract just the numeric values for the features and labels as the TensorFlow model will expect just numeric values as input."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Features Array shape:  (5000, 6)\n",
      "Label Array shape :  (5000,)\n"
     ]
    }
   ],
   "source": [
    "#Now, we will convert X and y into numpy arrays\n",
    "x_arr = X.values\n",
    "y_arr = y.values\n",
    "print('Features Array shape: ',x_arr.shape)\n",
    "print(\"Label Array shape : \",y_arr.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.4: Train and Validation Split\n",
    "\n",
    "We will keep some part of the data aside as a __validation__ set. The model will not use this set during training and it will be used only for checking the performance of the model in trained and un-trained states. This way, we can make sure that we are going in the right direction with our model training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Set Features:  (4750, 6)\n",
      "Training Set Labels:  (4750,)\n",
      "Validation Set Features:  (250, 6)\n",
      "Validation Set Labels:  (250,)\n"
     ]
    }
   ],
   "source": [
    "x_train,x_val,y_train,y_val = train_test_split(x_arr,y_arr,test_size = 0.05,\n",
    "                                               random_state = 0)\n",
    "print('Training Set Features: ', x_train.shape)\n",
    "print('Training Set Labels: ', y_train.shape)\n",
    "print('Validation Set Features: ', x_val.shape)\n",
    "print('Validation Set Labels: ', y_val.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create the Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5.1: Create the Model\n",
    "\n",
    "Let's write a function that returns an untrained model of a certain architecture."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 10)                70        \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 50)                550       \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 30)                1530      \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 1)                 31        \n",
      "=================================================================\n",
      "Total params: 2,181\n",
      "Trainable params: 2,181\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "def get_model():\n",
    "    model = Sequential([\n",
    "        Dense(10,input_shape = (6,),activation = 'relu'),\n",
    "        Dense(50,activation = 'relu'),\n",
    "        Dense(30, activation = 'relu'),\n",
    "        Dense(1) # As this is a regression problem, we don't use activation \n",
    "        #function at this layer\n",
    "    ])\n",
    "    \n",
    "    model.compile(\n",
    "        loss = 'mse',\n",
    "        optimizer = 'adam'\n",
    "    )\n",
    "    \n",
    "    return model\n",
    "\n",
    "get_model().summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Training"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6.1: Model Training\n",
    "\n",
    "We can use an `EarlyStopping` callback from Keras to stop the model training if the validation loss stops decreasing for a few epochs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 4750 samples, validate on 250 samples\n",
      "Epoch 1/500\n",
      "4750/4750 [==============================] - 1s 125us/sample - loss: 0.4397 - val_loss: 0.2235\n",
      "Epoch 2/500\n",
      "4750/4750 [==============================] - 0s 47us/sample - loss: 0.2047 - val_loss: 0.1786\n",
      "Epoch 3/500\n",
      "4750/4750 [==============================] - 0s 49us/sample - loss: 0.1804 - val_loss: 0.1638\n",
      "Epoch 4/500\n",
      "4750/4750 [==============================] - 0s 47us/sample - loss: 0.1713 - val_loss: 0.1617\n",
      "Epoch 5/500\n",
      "4750/4750 [==============================] - 0s 44us/sample - loss: 0.1667 - val_loss: 0.1622\n",
      "Epoch 6/500\n",
      "4750/4750 [==============================] - 0s 47us/sample - loss: 0.1640 - val_loss: 0.1627\n",
      "Epoch 7/500\n",
      "4750/4750 [==============================] - 0s 46us/sample - loss: 0.1614 - val_loss: 0.1603\n",
      "Epoch 8/500\n",
      "4750/4750 [==============================] - 0s 43us/sample - loss: 0.1598 - val_loss: 0.1645\n",
      "Epoch 9/500\n",
      "4750/4750 [==============================] - 0s 43us/sample - loss: 0.1588 - val_loss: 0.1565\n",
      "Epoch 10/500\n",
      "4750/4750 [==============================] - 0s 47us/sample - loss: 0.1578 - val_loss: 0.1565\n",
      "Epoch 11/500\n",
      "4750/4750 [==============================] - 0s 50us/sample - loss: 0.1574 - val_loss: 0.1586\n",
      "Epoch 12/500\n",
      "4750/4750 [==============================] - 0s 44us/sample - loss: 0.1552 - val_loss: 0.1609\n",
      "Epoch 13/500\n",
      "4750/4750 [==============================] - 0s 44us/sample - loss: 0.1548 - val_loss: 0.1577\n",
      "Epoch 14/500\n",
      "4750/4750 [==============================] - 0s 44us/sample - loss: 0.1548 - val_loss: 0.1612\n"
     ]
    }
   ],
   "source": [
    "es_cb = EarlyStopping(monitor = 'val_loss',patience = 5)\n",
    "\n",
    "model = get_model()\n",
    "preds_on_untrained = model.predict(x_val)\n",
    "\n",
    "history = model.fit(\n",
    "    x_train, y_train, \n",
    "    validation_data = (x_val,y_val),\n",
    "    epochs = 500,\n",
    "    callbacks = [es_cb]\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can note here, that our model has stopped at Epoch 14/500, which suggests that there was not considerable change in the validation loss value which is why the model stopped training and did not run on all the epochs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6.2: Plot Training and Validation Loss\n",
    "\n",
    "Let's use the `plot_loss` helper function to take a look training and validation loss."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_loss(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Predictions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 7.1: Plot Raw Predictions\n",
    "\n",
    "Let's use the `compare_predictions` helper function to compare predictions from the model when it was untrained and when it was trained."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "preds_on_trained = model.predict(x_val)\n",
    "\n",
    "compare_predictions(preds_on_untrained, preds_on_trained,y_val)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Insight : Untrained model makes random predictions which are all over the place,but it is a linear plot for the trained predictions. Surely, it has a high residual sum, but not as high as in the case of our untrained model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 7.2: Plot Price Predictions\n",
    "\n",
    "The plot for price predictions and raw predictions will look the same with just one difference: The x and y axis scale is changed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "price_untrained = [convert_label_value(y) for y in preds_on_untrained]\n",
    "price_trained = [convert_label_value(y) for y in preds_on_trained]\n",
    "price_val = [convert_label_value(y) for y in y_val]\n",
    "\n",
    "compare_predictions(price_untrained, price_trained,price_val)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Insight : We pretty much get the same graph, but the ranges are now different and correspond to the original housing prices"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}