Lasso Regression

Linearity

maintaining linearity with regularization that can shrink coefficients to zero

Outliers

Sensitive but capable of reducing some effects of outliers through coefficient shrinkage

Scalability

Computationally intensive

Multicollinearity

Handles multicollinearity and performs variable selection by setting some coefficients to zero

High dimensionality

Effective in high-dimensional settings by setting coefficients to zero

Interpretability

Somewhat interpretable through highlighting influential features

Ability to extrapolate

Capable of extrapolation within the range of the training data

Robustness

robustness similar to Ridge with additional benefits from feature selection

Assumptions of normality

Assumes normality of residuals