Lasso Regression

The least absolute shrinkage and selection operator (LASSO) regression is regression with \(L_1\) loss, $\( \mathcal L(\theta) = \sum_{i=1}^n L(y,\hat y) + \lambda\|\theta\|_1 \)$ The name is appropriate because the LASSO often sets a subset of the feature parameters to zero.