Online Program Home
My Program

Abstract Details

Activity Number: 124
Type: Contributed
Date/Time: Monday, August 1, 2016 : 8:30 AM to 10:20 AM
Sponsor: Section on Nonparametric Statistics
Abstract #319517 View Presentation
Title: Variable Bandwidth Local Polynomial Smoothing via Local Cross-Validation
Author(s): Katherine Grzesik* and Derick R. Peterson
Companies: University of Rochester and University of Rochester
Keywords: local polynomial ; cross-validation ; variable bandwidth

Nonparametrically estimating a regression function with varying degrees of smoothness or heteroscedasticity can benefit from a smoother that is more flexible than a constant bandwidth local polynomial to efficiently capture its features. We propose estimating a smooth variable bandwidth function using a form of Local Cross-Validation (LCV) based on smoothing the Squared Leave-One-Out cross-validated Residuals (SLOORs). Further, we propose a method of blending LCV fits of multiple polynomial orders, again guided by the smoothed SLOORs. The proposed method is computationally quicker than some current methods for functions with sharp changes in smoothness and can yield a reduction in mean squared error for moderate sample sizes while remaining competitive with larger samples.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2016 program

Copyright © American Statistical Association