Locally-weighted regression: An approach to regression analysis by local fitting.

Locally-weighted regression is a procedure for estimating a regression surface by a multivariate smoothing procedure: fitting a function of the independent variables locally and in a moving fashion that is analogous to how a moving average is computed for a time series. In using local fitting we can estimate a much wider class of regression surfaces than we can with the usual classes of parametric functions, such as polynomials. The goal of this paper is to show, through applications, how local regression can be used for three purposes: data exploration, regression diagnostics, and providing a regression surface.