Bandwidth selection in smoothing functions

Abstract

A simple criterion for selecting a bandwidth parameter that controls the amount of smoothing in functions is described. The procedure is computationally inexpensive and, hence, worth adopting. We argue that the bandwidth parameter is determined by two factors: the kernel function and the length of the smoothing region. We give an illustrative example of its application using real data.

Keywords: Kernel, Smoothing functions, Bandwidth

> East African Journal of Statistics Vol. 1 (2) 2006: pp. 162-174