An Error Analysis Technique for Statistical Hypothesis Testing. Generation of Jointly Distributed Random Variates,
Abstract
As a part of the study of electromagnetic scattering from rough terrain, various geographical sites were characterized by applying statistical analysis techniques to digitized terrain data bases. One aspect was the use of decision theory formulation to assign an appropriate distribution function to the surface height variations in the regions. This report discusses the decision processes used and the errors to be expected. There are two main themes: assessment of statistical hypothesis testing and application of a numerical technique to determine the errors inherent in the decisions. There are a number of assumptions and constraints that affect the form of the hypothesis test developed for the terrain study and these will be considered. The numerical technique consists in generating jointly distributed random variates in a Monte Carlo type computer procedure. Its application to the specification of decision errors will be described. The formulation allows extension of standard analytic error analysis techniques to cases where the analysis would be intractable. Results are presented and the agreement with analysis for various cases is shown. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1981
- Accession Number
- ADA103956
Entities
People
- John F. Lennon
- Robert J. Papa
Organizations
- Rome Laboratory