Optimal Quantile Principle for Selecting Variable Bandwidth in Regression Estimators

Andrzej S. Kozek and Eugene F. Schuster
Additional contact information
Andrzej S. Kozek: The University of Texas at El Paso, Department of Mathematical Sciences
Eugene F. Schuster: The University of Texas at El Paso, Department of Mathematical Sciences

A chapter in Computing Science and Statistics, 1992, pp 401-405 from Springer

Abstract: Abstract The optimal neighbor principle in data-driven selection of the variable window bandwidth in the Nadaraya-Watson regression estimators results in a lack of continuity of the bandwidth as a function of the conditional argument. This causes a lack of smoothness in the corresponding regression estimator which is frequently observed in samples of small or moderate size. A remedy, called the optimal quantile principle, is proposed and studied. We show that it is possible to modify the well-developed optimality theory used in choosing the smoothing parameter in Nadaraya-Watson kernel regression to give conditions under which corresponding optimality properties hold for the case of our p-th quantile estimator. Experience with computer simulations using a nonlinear regression package is reported.

Keywords: Regression Function; Kernel Estimator; Regression Estimator; Kernel Regression; Mean Integrate Square Error (search for similar items in EconPapers)
Date: 1992
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:sprchp:978-1-4612-2856-1_63

Ordering information: This item can be ordered from
http://www.springer.com/9781461228561

DOI: 10.1007/978-1-4612-2856-1_63

Access Statistics for this chapter

More chapters in Springer Books from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().