Nonparametric regression analyzes how a response changes based on predictors without assuming a specific relationship. Common methods include kernel estimation, local polynomial regression, and smoothing splines. Additive regression and semiparametric models are also used.