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 #319626
Title: Convolution Weighting in Compound Estimation with or Without Heteroscedasticity
Author(s): Sisheng Liu* and Richard Charnigo and Cidambi Srinivasan
Companies: University of Kentucky and University of Kentucky and University of Kentucky
Keywords: Nonparametric regression ; Compound estimation ; heteroscedasticity
Abstract:

Compound estimator (Charnigo et al 2011; Charnigo et al 2015) has been developed for simultaneously estimating the mean response function and its derivatives. When trying to address heteroscedasticity in compound estimation, a Cp criterion is employed for choosing parameters - including local bandwidths and convolution weights. Our framework also differs from the original version of compound estimation in that the convolution weights replace a single convolution parameter. The essentially optimal convergence rate in the original version still holds with this modification. Moreover, the Cp criterion will minimize a discretized approximation to Integrated Mean Square Error to choose the parameters. Also, the simulation studies showed that choosing bandwidth and convolution weights sequentially works well.


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

Back to the full JSM 2016 program

 
 
Copyright © American Statistical Association