Ridge Regression

Linear regression with L2 regularization to prevent overfitting and handle multicollinearity.

When to use:

• Have correlated features (multicollinearity)
• More features than samples
• Want to prevent overfitting
• Linear relationships but need regularization

Strengths: Handles multicollinearity well, prevents overfitting, all features kept (none zeroed out) Weaknesses: Doesn't perform feature selection, still assumes linearity

Model Parameters

Alpha (default: 1.0) Regularization strength. Higher values = stronger regularization = simpler model.

• Low (0.01-0.1): Weak regularization, close to Linear Regression
• Default (1.0): Balanced
• High (10-100): Strong regularization, very simple model

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

Solver

• auto: Choose automatically (default)
• svd: Singular value decomposition (most stable)
• cholesky: Fast for many features
• lsqr: Good for large problems
• saga: Fast for large datasets