Created
February 20, 2020 22:10
-
-
Save SimonAB/60645eea9cb32ee714635e36a584e262 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
""" | |
This function uses gradient descent to search for the weights | |
that minimises the logit cost function. | |
A tuple with learned weights vector (θ) and the cost vector (𝐉) | |
are returned. | |
""" | |
function logistic_regression_sgd(X, y, λ, fit_intercept=true, η=0.01, max_iter=1000) | |
# Initialize some useful values | |
m = length(y); # number of training examples | |
if fit_intercept | |
# Add a constant of 1s if fit_intercept is specified | |
constant = ones(m, 1) | |
X = hcat(constant, X) | |
else | |
X # Assume user added constants | |
end | |
# Use the number of features to initialise the theta θ vector | |
n = size(X)[2] | |
θ = zeros(n) | |
# Initialise the cost vector based on the number of iterations | |
𝐉 = zeros(max_iter) | |
for iter in range(1, stop=max_iter) | |
# Calcaluate the cost and gradient (∇𝐉) for each iter | |
𝐉[iter], ∇𝐉 = regularised_cost(X, y, θ, λ) | |
# Update θ using gradients (∇𝐉) for direction and (η) for the magnitude of steps in that direction | |
θ = θ - (η * ∇𝐉) | |
end | |
return (θ, 𝐉) | |
end |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment