| Simple Linear Regression |
One independent variable, one dependent variable |
y = β₀ + β₁·x + ε |
Predicting house price based on square footage |
| Multiple Linear Regression |
Multiple independent variables |
y = β₀ + β₁·x₁ + β₂·x₂ + ... + βₙ·xₙ + ε |
Forecasting sales revenue based on marketing metrics |
| Polynomial Regression |
Fits a nonlinear curve using polynomial terms |
y = β₀ + β₁·x + β₂·x² + ... |
Modeling population growth over time |
| Ridge Regression |
Prevents overfitting by shrinking coefficients (L2) |
Minimize: ∑(yᵢ - ŷᵢ)² + α∑βⱼ² |
Stock prediction with many similar features |
| Lasso Regression |
Performs feature selection by shrinking some coefficients to 0 (L1) |
Minimize: ∑(yᵢ - ŷᵢ)² + α∑|βⱼ| |
Detecting major factors in customer churn |
| Elastic Net |
Combines Ridge and Lasso for balanced regularization |
Minimize: ∑(yᵢ - ŷᵢ)² + α₁∑|βⱼ| + α₂∑βⱼ² |
Analyzing genetics with correlated predictors |
No comments:
Post a Comment