Variable selection in finite mixture of generalized estimating equations. Issue 6 (3rd June 2022)

• Record Type: Journal Article
• Authors: Rezazadeh, H.; Eskandari, F.; Moghadam, M. Bameni; Ormoz, E.
• Is Part Of: Communications in statistics. Volume 51:Issue 6(2022)
• Page Start: 3237
• Page End: 3251
• Publication Date: 2022-06-03
• Subjects: Iterative Weighted Least Square -- Least Square Principle -- Mixture model -- Variable selection; Mathematical statistics -- Periodicals; Mathematical statistics -- Data processing -- Periodicals; Digital computer simulation -- Periodicals; 519.5
• DOI: 10.1080/03610918.2019.1711406
• Languages: English
• ISSNs: 0361-0918

Abstract

Abstract: This paper develops a new method to estimate the parameters in mixture models. Traditionally, the parameter estimation in mixture models is performed from a likelihood point of view by exploiting the expectation maximization (EM) method. In this paper, however, we utilize the Least Square Principle. Based on this principle, we propose an iterative algorithm called Iterative Weighted least Square (IWLS) to estimate the parameters. Through comparative study, we demonstrate the superiority of our method compared to EM method. We show that IWLS method outperforms EM in both accuracy and the number of iterations required for convergence.