CompTIA DataX DY0-001 (V1) Practice Question

The correct answer is Elastic Net regression. This model is specifically designed to handle situations with a high number of features and multicollinearity.

• Elastic Net combines the L1 (Lasso) and L2 (Ridge) penalties. The L1 penalty performs feature selection by shrinking some coefficients to exactly zero, creating a parsimonious model. The L2 penalty effectively handles multicollinearity by shrinking the coefficients of correlated predictors together, rather than arbitrarily selecting one. This dual approach directly addresses both requirements outlined in the scenario.
• LASSO Regression uses an L1 penalty and is effective for feature selection. However, when faced with a group of highly correlated variables, it tends to arbitrarily select only one and shrink the others to zero, which can lead to model instability and loss of information.
• Ridge Regression uses an L2 penalty and is excellent for managing multicollinearity. However, it does not perform feature selection; it shrinks coefficients towards zero but will not set them exactly to zero, meaning all features remain in the model.
• Ordinary Least Squares (OLS) Regression does not use any regularization. It is highly sensitive to multicollinearity, which can lead to unstable and unreliable coefficient estimates, and it does not perform any feature selection, making it prone to overfitting with a large number of features.