ON THE USE OF TWO-SAMPLE DECISION PROCEDURE FOR DEPENDENT RANDOM VARIABLES AS A TEST OF STRICT-SENSE STATIONARITY.

Abstract

This is an expository report of some significant results obtained by K. Matusita in the treatment of the classical 'two sample problem' in statistics, when the random variables involved are not necessarily statistically independent. These resuls should be particularly useful in testing the hypothesis of strict-sense stationarity for a sample of radar data obtained from a single sample function over a finite interval in time (O,T). For example, if the interval (O,T) is divided into n equal subintervals, then data samples from pairs of subintervals may be tested to see if they are from the same distribution. The essence of Matusita's results are presented with some extensions and additional details where required in the interest of clarity, and they are applied as a distributionfree test of the hypothesis of strict-sense stationarity for a generalized random process. (Author)

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 1965
Accession Number
AD0467488

Entities

People

  • G. W. Evans Ii.
  • Rose McCarty

Organizations

  • SRI International

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Data Science
  • Information Science
  • Intervals
  • Mathematics
  • Random Variables
  • Statistics

Fields of Study

  • Mathematics

Readers

  • Regression Analysis.
  • Statistical inference.