"""
Training a linear regression model using Keras
"""
# Import required packages:
from keras.models import Sequential
from keras.layers import Dense
from keras.optimizers import Adam
import numpy as np
import matplotlib.pyplot as plt
# Number of points:
N = 50
# Make random numbers predictable:
np.random.seed(101)
# Generate random data composed by 50 (N = 50) points:
x = np.linspace(0, N, N)
y = 3 * np.linspace(0, N, N) + np.random.uniform(-10, 10, N)
def get_weights(model):
"""Get weights of w and b"""
w = model.get_weights()[0][0][0]
b = model.get_weights()[1][0]
return w, b
def create_model():
"""Create the model using Sequential model"""
# Create a sequential model:
model = Sequential()
# All we need is a single connection so we use a Dense layer with linear activation:
model.add(Dense(input_dim=1, units=1, activation="linear", kernel_initializer="uniform"))
# Compile the model defining mean squared error(mse) as the loss
model.compile(optimizer=Adam(lr=0.1), loss='mse')
# Return the created model
return model
# Get the created model:
linear_reg_model = create_model()
# Show weights at beginning (initialization values):
w_init, b_init = get_weights(linear_reg_model)
print('Linear regression model is initialized with weights w: {}, b: {}'.format(w_init, b_init))
# Feed the data using fit function:
linear_reg_model.fit(x, y, epochs=100, validation_split=0.2, verbose=1)
# Show weights when the training is done (learned parameters):
w_final, b_final = get_weights(linear_reg_model)
print('Linear regression model is trained with weights w: {}, b: {}'.format(w_final, b_final))
# Calculate the predictions:
predictions = w_final * x + b_final
# Saving weights:
linear_reg_model.save_weights("my_model.h5")
# Create the dimensions of the figure and set title:
fig = plt.figure(figsize=(8, 5))
plt.suptitle("Linear regression using Keras", fontsize=14, fontweight='bold')
fig.patch.set_facecolor('silver')
# Plot training data:
plt.subplot(1, 2, 1)
plt.plot(x, y, 'ro', label='Original data')
plt.xlabel('x')
plt.ylabel('y')
plt.title("Training Data")
# Plot results:
plt.subplot(1, 2, 2)
plt.plot(x, y, 'ro', label='Original data')
plt.plot(x, predictions, label='Fitted line')
plt.xlabel('x')
plt.ylabel('y')
plt.title('Linear Regression Result')
plt.legend()
# Show the Figure:
plt.show()
