Elastic Net is a regularized regression technique that linearly combines L1 (Lasso) and L2 (Ridge) penalties on model coefficients. In a regression setting, the elastic net optimization problem adds a penalty term αL1 + βL2 to the loss function (with α and β controlling the mix of Lasso vs Ridge). The effect is that elastic net can perform feature selection (like Lasso, driving some coefficients to zero) while also mitigating some limitations of Lasso when variables are highly correlated (where Ridge’s grouping effect helps). It thus inherits the strengths of both: L1 aids sparse model interpretation, and L2 stabilizes the solution and can handle collinearity. The net result is often a more robust model than either Lasso or Ridge alone, especially in high-dimensional data where p >> n. The regularization hyperparameters are usually tuned via cross-validation.