Kernel-Type Estimators of Jump Points and Values of a Regression Function

Abstract

The authors propose kernel type estimators of the locations of jump points and the corresponding sizes of jumps of the regression function. They study their limit properties and demonstrate their performance on simulations.