On Jackknifing Kernel Regression Function Estimators.
Abstract
Estimation of the value of a regression function at a point of continuity using a kernal-type estimator is discussed and improvements of the technique by a generalized jackknife estimator are presented. It is shown that the generalized jackknife technique produces estimators with faster bias rates. In a small example it is investigated, if the generalized jackknife method works for all choices of bandwidths. It turns out that an improper choice of this parameter may inflate the mean square error of the generalized jackknife estimator. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1983
- Accession Number
- ADA133236
Entities
People
- Wolfgang Hardle
Organizations
- University of North Carolina at Chapel Hill