Empirical Quantile Function Nonparametric Integrated Analyses
Abstract
This three-year statistics research project addressed theoretical development, efficiency evaluation, and desktop computer implementation of an integrated nonparametric statistical data analysis approach based on Fourier analytic techniques applied to the empirical quantile function (EQF). New EQF statistical procedures were developed for data smoothing and reduction methods, nonparametric estimation of various functionals associated with the quantile function, composite goodness-of-fit tests for uniform and exponential models with applications to related stochastic processes, and nonparametric analysis of variance rank procedures for analysis of independent samples. These EQF methods exploit discrete Hahn polynomial orthogonal polynomial component representations to produce statistics for significance testing and interval estimation for both omnibus and directional alternative models. Performance evaluations of the proposed techniques included theoretical and Monte Carlo efficiency comparisons as well as numerical applications to challenging data sets. Interactive desktop computer and graphics routines utilizing symbolic programming languages were developed to facilitate implementation of EQF data analysis techniques by statistical users. Nonparametric Statistics, Quantile Function, Goodness of Fit, Components.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1993
- Accession Number
- ADA263704
Entities
People
- W. D. Kaigh
Organizations
- University of Texas at El Paso