Nonparametric Selection, Ranking and Testing
Abstract
In this paper we consider the problem of ranking (partitioning) k populations according to the parameter which is defined as functionals of the distribution functions on the underlying populations. We obtain minimax rules for general loss functions, Bayes rules for some specific loss functions and purpose approximate non-randomized minimax rules. We also derive restricted minimax rules for selecting a subset of populations which are better than a control. Some nonparametric optimal test are derived for different hypotheses written in terms of the parameter as a functional of the underlying distribution function. Keywords: Selection and ranking; Nonparametric; Comparison with a control; Testing; Minimax decision rules; Mathematics.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1990
- Accession Number
- ADA219254
Entities
People
- Sayaji N. Hande
Organizations
- Purdue University