Robust Estimation Techniques for Location Parameter Estimation of Symmetric Distributions
AIR FORCE INST OF TECH WRIGHT-PATTERSON AFB OH SCHOOL OF ENGINEERING
Pagination or Media Count:
Several robust estimators are considered for analysis and explanation. Monte Carlo techniques are used to investigate the efficiency of these robust estimators relative to the best estimator for the distribution under consideration. Sample sizes of 12 and 24 were drawn 4200 times from five symmetric probability distributions. The results show that over a class of distributions the robust estimators provide a higher guaranteed efficiency than the best estimator for any particular distribution in the family. Some interesting results are apparent from an analysis of the graphs in Appendix C indicating some upper bounds on the size of the Monte Carlo sample when conducting this type of study.
- Statistics and Probability