Keras Deep Neural Net Template #python #ml #tensorflow

KerasDeepNeuralNet.py

# Setup plotting
import matplotlib.pyplot as plt
from learntools.deep_learning_intro.dltools import animate_sgd
plt.style.use('seaborn-whitegrid')
# Set Matplotlib defaults
plt.rc('figure', autolayout=True)
plt.rc('axes', labelweight='bold', labelsize='large',
       titleweight='bold', titlesize=18, titlepad=10)
plt.rc('animation', html='html5')


## Preprocessing
import numpy as np
import pandas as pd
from sklearn.preprocessing import StandardScaler, OneHotEncoder
from sklearn.compose import make_column_transformer, make_column_selector
from sklearn.model_selection import train_test_split

fuel = pd.read_csv('../input/dl-course-data/fuel.csv')

X = fuel.copy()
# Remove target
y = X.pop('FE')

preprocessor = make_column_transformer(
    (StandardScaler(),
     make_column_selector(dtype_include=np.number)),
    (OneHotEncoder(sparse=False),
     make_column_selector(dtype_include=object)),
)

X = preprocessor.fit_transform(X)
y = np.log(y) # log transform target instead of standardizing

input_shape = [X.shape[1]]
print("Input shape: {}".format(input_shape))


## Look at Data
#fuel.head()
# Uncomment to see processed features
pd.DataFrame(X[:10,:]).head()


## Modelling
from tensorflow import keras
from tensorflow.keras import layers

model = keras.Sequential([
    layers.Dense(128, activation='relu', input_shape=input_shape),
    layers.Dense(128, activation='relu'),    
    layers.Dense(64, activation='relu'),
    layers.Dense(1),
])


## Compile
model.compile(
    optimizer='adam',
    loss='mae',
)


## Train
history = model.fit(
    X_train, y_train,
    validation_data=(X_valid, y_valid),
    batch_size=256,
    epochs=10,
)


## Plot Loss
import pandas as pd

# convert the training history to a dataframe
history_df = pd.DataFrame(history.history)
# use Pandas native plot method
history_df['loss'].plot();