A linear regression model has many features, and the data scientist suspects most of them are irrelevant. They want a regularization technique that automatically drives irrelevant feature coefficients to exactly zero, performing feature selection. Which regularization method achieves this?

Question

Accepted Answer

B. L1 regularization (Lasso). L1 regularization (Lasso) adds a penalty proportional to the absolute value of coefficients, which creates a sparse solution by driving irrelevant feature coefficients to exactly zero — effectively performing automatic feature selection.

A linear regression model has many features, and the data scientist suspects most of them are irrelevant. They want a regularization technique that automatically drives irrelevant feature coefficients to exactly zero, performing feature selection. Which regularization method achieves this?

Related Questions

Discussion