© 1985 by Biometrika Trust
MISEALLANEA |
Asymptotic nonequivalence of some bandwidth selectors in nonparametric regression
Fachbereich Mathematik, Johann Wolfgang Goethe Universität D-6000 Frankfurt am Main, Federal Republic of Germany
Department of Statistics, University of North Carolina Chapel Hill, N.C. 27514, U.S.A.
SUMMARY
The bandwidth selection problem in nonparametric kernel regression is considered. Bandwidth selectors based on cross-validation and on Akaike's information criterion, atc, and his finite prediction error, FBE, are among those compared. It is seen that they are not necessarily asymptotically equivalent. Conditions are given under which the equivalence holds and modifications are suggested which make the selectors equivalent.
Key Words: Bandwidth selection Kernel estimator Model selection Nonparametric regression