Last active
March 15, 2020 12:48
-
-
Save FisherKK/a8032a8d64effb3d676c053f5fbf96f7 to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
def train(X, y, model_parameters, learning_rate=0.0005, iterations=20000): | |
# Make prediction for every data sample | |
predictions = [predict(x, model_parameters) for x in X] | |
# Calculate initial cost for model - MSE | |
initial_error = mse(predictions, y) | |
print("Initial state:") | |
print(" - error: {}".format(initial_error)) | |
print(" - parameters: {}".format(model_parameters)) | |
for i in range(iterations): | |
# Sum up partial gradients for every data sample, for every parameter in model | |
accumulated_grad_w0 = 0 | |
accumulated_grad_b = 0 | |
for x, y_target in zip(X, y): | |
accumulated_grad_w0 += (predict(x, model_parameters) - y_target)*x[0] | |
accumulated_grad_b += (predict(x, model_parameters) - y_target) | |
# Calculate mean of gradient | |
w_grad = (1.0/len(X)) * accumulated_grad_w0 | |
b_grad = (1.0/len(X)) * accumulated_grad_b | |
# Update parameters by small part of averaged gradient | |
model_parameters["w"][0] = model_parameters["w"][0] - learning_rate * w_grad | |
model_parameters["b"] = model_parameters["b"] - learning_rate * b_grad | |
if i % 4000 == 0: | |
print("\nIteration {}:".format(i)) | |
print(" - error: {}".format(mse([predict(x, model_parameters) for x in X], y))) | |
print(" - parameters: {}".format(model_parameters)) | |
print("\nFinal state:") | |
print(" - error: {}".format(mse([predict(x, model_parameters) for x in X], y))) | |
print(" - parameters: {}".format(model_parameters)) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hi
Is it possible to get the csv data file? thx