Skip to content

IAI Actuarial Core Principles · Risk Modelling and Survival Analysis · Elementary principles of machine learning

A lasso (L1-penalised) regression is fitted to predict lapse rates and the penalty parameter lambda is increased from a small value to a very large value. Which outcome is expected?

As lambda rises, the lasso shrinks coefficients and sets more of them exactly to zero, giving a simpler model. This reduces variance but increases bias, and training error rises rather than falls.

  1. AMore coefficients are shrunk to exactly zero and the model's variance falls while bias risesCorrect
  2. BAll coefficients increase in magnitude and bias falls
  3. CTraining error falls steadily as lambda increases
  4. DCoefficients are shrunk but never become exactly zero
  5. The model becomes more flexible and overfits more

Explanation

Increasing the lasso penalty shrinks coefficients, with some set exactly to zero, reducing model complexity. This lowers variance but increases bias, and training error generally rises. Never reaching exactly zero describes ridge regression, not lasso.

Did you get it right without looking?

One question tells you little. A timed set on Elementary principles of machine learning shows your real accuracy, how long you take and where you lose marks.

Practise Elementary principles of machine learningFree diagnostic mock

More Elementary principles of machine learning questions