Login to manage your account

Your email Please enter a valid email address.

Password Forgot Password?

Please enter a valid password.

OR

Don't have an account yet? Sign up

Why must features be scaled before applying L1 or L2 regularisation?

A The penalty depends on coefficient magnitude, so unscaled features are penalised unequally
B Regularisation only works on positive values
C Scaling is required by all models
D It reduces memory use

Answer

The penalty depends on coefficient magnitude, so unscaled features are penalised unequally

Tree-based models are scale-invariant and do not require it, but distance-based and gradient-descent models generally do.

All Machine Learning MCQs

Login to manage your account

Your email id Please enter a valid email address.

Password Forgot Password?

Please enter a valid password.

OR

Don't have an account yet? Sign up as

Jobseeker

Employer