Nonlinear regression model checking via local polynomial smoothing

Abstract

A goodness-of-fitt test statistic for nonlinear regression models based on local polynomial estimation is proposed in this paper. The criterion used to construct the test is the distance between the parametric fitt and the nonparametric regression estimation. The good performance of the test is shown via a simulation study. The method is applied to check a logistic mixture regression model for real data coming from a thermal analysis problem.