FUN.STAT Quantile Approach to Two Sample Statistical Data Analysis.
Abstract
FUN.STAT is a name proposed to describe a synthesis of statistical reasoning which combines quantiles and quantile-densities, information and entropy, and functional statistical inference. This paper describes a FUN.STAT approach to the problem of statistical data analysis of two random samples, respectively representing two populations of interest. It is composed of four parts. Part 1 describes how conventional approaches to two sample problems, including representations of linear rank statistics, are equivalent to functionals of a stochastic process. Part 2 motivates the autoregressive density estimation approach to the problem of functional statistical inference of this stochastic process and states several conjectures concerning the properties of the density estimation approach. Part 3 outlines heuristic derivations of the asymptotic distribution theory of the process. Part 4 provides a summary and an example, using TWOSAM which is a computer program for autoregressive two sample statistical data analysis; it has been implemented as a Fortran program and as a SAS procedure.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1983
- Accession Number
- ADA127672
Entities
People
- Emanuel Parzen
Organizations
- Texas A&M University