Accession Number : AD0777865

Title :   Robust Estimaton Techniques for Population Parameters and Regression Coefficients.

Descriptive Note : Master's thesis,

Corporate Author : AIR FORCE INST OF TECH WRIGHT-PATTERSON AFB OHIO SCHOOL OF ENGINEERING

Personal Author(s) : Forth,Charles R.

Report Date : MAR 1974

Pagination or Media Count : 67

Abstract : A Monte Carlo analysis is performed to determine the efficiencies of robust estimators of the location and scale parameters of the double exponential, normal, and uniform distributions relative to the maximum-likelihood estimators of the same parameters. Tables are prepared comparing the relative efficiencies of ten estimators of the location parameter and seven estimators of the scale parameter for each of the three distribution functions mentioned above. Besides the relative efficiencies, two other measures of merit are used to compare the robust estimators. These measures of merit are guaranteed efficiency and average efficiency. The Monte Carlo analysis also includes a procedure to determine the coefficients in a regression model by robust methods. The kurtosis of the residuals for the least squares regression is used as a discriminant for selecting the best method of regression. (Modified author abstract)

Descriptors :   *Distribution functions, *Regression analysis, Sampling, Probability density functions, Estimates, Tables(Data), Theses

Subject Categories : Statistics and Probability

Distribution Statement : APPROVED FOR PUBLIC RELEASE