Supervised Learning: Regression

Linear regression, regularization, and regression techniques.

Extending linear regression to non-normal responses via link functions – unifying logistic, Poisson, and other regression types.

Fitting a hyperplane to data by minimizing squared errors – the most interpretable and foundational predictive model.

Capturing nonlinear relationships within the linear regression framework by adding polynomial feature terms.

Residual analysis, heteroscedasticity, multicollinearity, and influence points – verifying assumptions before trusting results.

L2 and L1 penalties that shrink coefficients toward zero – Ridge for stability, Lasso for sparsity and feature selection.