Asymptotic Properties of Classification Rules Based on Wilcoxon-Type Statistics
研究了基于Wilcoxon型统计量的非参数分类规则在位置参数总体中的渐近性质，并通过误分类概率的渐近展开比较其与最优参数规则的效率。
Abstract Suppose training samples are available from two location parameter populations. Nonparametric classification rules based on Wilcoxon-type statistics are defined. The efficiencies of the nonparametric rules relative to the “optimal” estimated parametric rules are then investigated by using asymptotic expansions of the probabilities of misclassification.