The next meeting of the

National Seminar on Probability and Statistics

will be held on June 4, 2024, at 2:00 p.m. in Room 503 of the Institute of Mathematics and Informatics.

A talk on:

Comparison of parameter estimation methods for a Beta-Uniform mixture model

will be delivered by

Nikolay I. Nikolov (IMI – BAS).

Abstract.The Benjamini-Hochberg (B-H) method is a procedure for controlling the false discovery rate in the problem of multiple comparisons and is one of the most cited scientific works. In this talk, a Beta-Uniform mixture is considered for approximating the distribution of the p-values before the B-H adjustment. Expectation-maximization algorithms are derived for finding the maximum-likelihood and the maximum product of spacings estimations of the parameters in the suggested model. In addition, the method of moments is also applied for estimating the unknown parameters. The bias and mean squared error of the proposed estimation procedures are compared via numerical simulations. The talk is based on a joint work with Dean Palejev (Bulgarian Academy of Sciences, Sofia University).