×

Nearest neighbor rules for statistical classification based on ranks. (English) Zbl 0552.62044

This paper presents the problem of classifying an observation into one of two populations from which random training samples are available, and classification rules based on ”rank nearest neighbors”, a nonparametric procedure for statistical classification. Asymptotic values of the probabilities of misclassification of these rules are derived, and simple estimates based on runs are given in order to estimate the probabilities of misclassification. Their asymptotic properties are also discussed.
Reviewer: H.Bozdogan

MSC:

62H30 Classification and discrimination; cluster analysis (statistical aspects)
62G99 Nonparametric inference
PDFBibTeX XMLCite