ElasticNet Regression

Combines L1 (Lasso) and L2 (Ridge) regularization for best of both worlds.

When to use:

• Have grouped correlated features
• Want some feature selection but not too aggressive
• More stable than pure Lasso
• Best all-around regularized linear model

Strengths: Balances feature selection and multicollinearity handling, more stable than Lasso, flexible Weaknesses: Two hyperparameters to tune (alpha and l1_ratio)

Model Parameters

Alpha (default: 1.0) Overall regularization strength.

L1 Ratio (default: 0.5) Mix of L1 and L2 penalties:

• 0: Pure Ridge (L2)
• 0.5: Equal mix (default)
• 1: Pure Lasso (L1)

Max Iterations (default: 1000) Maximum optimization iterations.

Fit Intercept (default: true) Whether to calculate intercept term.