Statistics

Regularisation

Also known as: Ridge, Lasso, Shrinkage

Penalising model complexity to reduce variance at the cost of some bias.

Ridge (L2) shrinks coefficients smoothly toward zero and handles multicollinearity well. Lasso (L1) shrinks and sets some coefficients exactly to zero, giving feature selection. Elastic net combines both.

Why L1 zeroes coefficients and L2 does not is a standard follow-up: the L1 penalty has a corner at zero, so the constrained optimum often lands exactly on an axis.

Practical note. Standardise features first, or the penalty applies unevenly across variables measured on different scales - a common silent error.

Related terms

Practise this

Put it into practice

Knowing the definition is not the same as spotting where it applies under time pressure. Work the question bank free.

Start practising free

Browse the full quant interview glossary