explore_l2_regularization.py

# generate 100 different models with different lambda values, from 0 to 50
lambdas = np.linspace(0, 50, 200)
bias = []
variance = []
for λ in lambdas:
    
    # refit model with new λ hyperparameter
    β = np.linalg.inv(X.T.dot(X) + np.diag([λ] * len(X.T))).dot(X.T).dot(y).flatten()
    bias.append(np.mean(np.mean(np.abs(X.dot(β) - y.values.flatten()))))
    variance.append(np.var(X.dot(β)))