NN/Numerical Regressor: Vehicle Mileage Prediction¶

In [ ]:

import pandas as pd
import sklearn

import pandas as pd import sklearn

Pandas is most likeley used when missing data / fixing data

In [ ]:

url = 'http://archive.ics.uci.edu/ml/machine-learning-databases/auto-mpg/auto-mpg.data'
column_names = ['MPG', 'Cylinders', 'Displacement', 'Horsepower', 'Weight',
                'Acceleration', 'Model Year', 'Origin']

df = pd.read_csv(url, names=column_names,
                          na_values='?', comment='\t',
                          sep=' ', skipinitialspace=True)

url = 'http://archive.ics.uci.edu/ml/machine-learning-databases/auto-mpg/auto-mpg.data' column_names = ['MPG', 'Cylinders', 'Displacement', 'Horsepower', 'Weight', 'Acceleration', 'Model Year', 'Origin'] df = pd.read_csv(url, names=column_names, na_values='?', comment='\t', sep=' ', skipinitialspace=True)

In [ ]:

df.isnull().sum()

df.isnull().sum()

Out[ ]:

MPG             0
Cylinders       0
Displacement    0
Horsepower      6
Weight          0
Acceleration    0
Model Year      0
Origin          0
dtype: int64

In [ ]:

df

df

Out[ ]:

	MPG	Cylinders	Displacement	Horsepower	Weight	Acceleration	Model Year	Origin
0	18.0	8	307.0	130.0	3504.0	12.0	70	1
1	15.0	8	350.0	165.0	3693.0	11.5	70	1
2	18.0	8	318.0	150.0	3436.0	11.0	70	1
3	16.0	8	304.0	150.0	3433.0	12.0	70	1
4	17.0	8	302.0	140.0	3449.0	10.5	70	1
...	...	...	...	...	...	...	...	...
393	27.0	4	140.0	86.0	2790.0	15.6	82	1
394	44.0	4	97.0	52.0	2130.0	24.6	82	2
395	32.0	4	135.0	84.0	2295.0	11.6	82	1
396	28.0	4	120.0	79.0	2625.0	18.6	82	1
397	31.0	4	119.0	82.0	2720.0	19.4	82	1

398 rows × 8 columns

In [ ]:

df["MPG"].mean()

df["MPG"].mean()

Out[ ]:

23.514572864321607

In [ ]:

df = df.dropna()

df = df.dropna()

In [ ]:

# columns = ["Cylinders", "Model Year", "Origin"]
df = df.astype(float)

# columns = ["Cylinders", "Model Year", "Origin"] df = df.astype(float)

Split data into 3 parts

• Train
• Validation
• Test

We are here merging the validation and test due to lack to data entries

In [ ]:

from  sklearn.model_selection import train_test_split
import numpy as np

X = df.drop('MPG', axis=1) # Capital because that is a matrix -> Input Matrix
y = df['MPG'] # Lowercase since the data is an array -> Output Array

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2)
# 0.2 -> Split ration between train and test

X_train.shape, X_test.shape, y_train.shape, y_test.shape
type(y_train)

from sklearn.model_selection import train_test_split import numpy as np X = df.drop('MPG', axis=1) # Capital because that is a matrix -> Input Matrix y = df['MPG'] # Lowercase since the data is an array -> Output Array X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2) # 0.2 -> Split ration between train and test X_train.shape, X_test.shape, y_train.shape, y_test.shape type(y_train)

Out[ ]:

pandas.core.series.Series

In [ ]:

from tensorflow.keras.layers import Dense, SimpleRNN
import tensorflow as tf

from tensorflow.keras.layers import Dense, SimpleRNN import tensorflow as tf

In [ ]:

my_model = tf.keras.Sequential([
    # Input((1, 8))
    Dense(100, activation="relu"),
    # 100 * (8+1)
    Dense(10, activation="relu"),
    SimpleRNN(units=1, activation='tanh', name='Hidden-Recurrent-Layer'))
    # 10*(100+1)
    Dense(10, activation="relu"),
    # 10*(10+1)
    Dense(1)
    # 1*(10+1)
])

my_model.compile(loss = tf.keras.losses.MeanSquaredError(), optimizer=tf.keras.optimizers.Adam(), metrics=["mae"])
# accuraccy -> If data is matching
# MSE, MSLE, MAE -> Difference btw prediction and expected
model_history = my_model.fit(X_train, y_train, epochs=10, validation_data=(X_test, y_test))

my_model = tf.keras.Sequential([ # Input((1, 8)) Dense(100, activation="relu"), # 100 * (8+1) Dense(10, activation="relu"), SimpleRNN(units=1, activation='tanh', name='Hidden-Recurrent-Layer')) # 10*(100+1) Dense(10, activation="relu"), # 10*(10+1) Dense(1) # 1*(10+1) ]) my_model.compile(loss = tf.keras.losses.MeanSquaredError(), optimizer=tf.keras.optimizers.Adam(), metrics=["mae"]) # accuraccy -> If data is matching # MSE, MSLE, MAE -> Difference btw prediction and expected model_history = my_model.fit(X_train, y_train, epochs=10, validation_data=(X_test, y_test))

Epoch 1/10
10/10 [==============================] - 9s 45ms/step - loss: 17373.5020 - mae: 121.5429 - val_loss: 1365.5695 - val_mae: 35.7540
Epoch 2/10
10/10 [==============================] - 0s 13ms/step - loss: 698.9968 - mae: 24.9100 - val_loss: 587.5173 - val_mae: 23.0756
Epoch 3/10
10/10 [==============================] - 0s 23ms/step - loss: 377.1506 - mae: 16.9788 - val_loss: 125.7078 - val_mae: 9.3703
Epoch 4/10
10/10 [==============================] - 0s 17ms/step - loss: 180.9805 - mae: 11.3088 - val_loss: 140.1292 - val_mae: 9.8517
Epoch 5/10
10/10 [==============================] - 0s 19ms/step - loss: 151.1945 - mae: 10.1285 - val_loss: 139.6962 - val_mae: 9.6343
Epoch 6/10
10/10 [==============================] - 0s 15ms/step - loss: 149.4257 - mae: 9.9768 - val_loss: 124.9488 - val_mae: 9.3445
Epoch 7/10
10/10 [==============================] - 0s 15ms/step - loss: 142.7760 - mae: 9.9751 - val_loss: 122.1221 - val_mae: 9.2163
Epoch 8/10
10/10 [==============================] - 0s 11ms/step - loss: 139.6175 - mae: 9.7384 - val_loss: 119.7234 - val_mae: 9.0681
Epoch 9/10
10/10 [==============================] - 0s 11ms/step - loss: 135.7001 - mae: 9.5561 - val_loss: 116.0868 - val_mae: 8.9237
Epoch 10/10
10/10 [==============================] - 0s 16ms/step - loss: 134.3057 - mae: 9.5375 - val_loss: 113.3275 - val_mae: 8.7397

val_mae 8.7 meaning the error is close to 8.7 miles per gallon when taking into consideration so many cars, this translates to model being very close to accurate

In [ ]:

pd.DataFrame(model_history.history).plot()

pd.DataFrame(model_history.history).plot()

Out[ ]:

<Axes: >