Estimation of the mixtures of GLMs with covariate-dependent mixing proportions. Issue 24 (16th December 2016)

• Record Type: Journal Article
• Authors: Wu, Xing; Yu, Conglian
• Is Part Of: Communications in statistics. Volume 45:Issue 24(2016)
• Page Start: 7242
• Page End: 7257
• Publication Date: 2016-12-16
• Subjects: EM algorithm -- Generalized linear models -- Kernel regression -- Mixture of regression models.; 62J12; Mathematical statistics -- Periodicals; Mathematics; Statistics; 519.2
• DOI: 10.1080/03610926.2014.975822
• Languages: English
• ISSNs: 0361-0926

Abstract

ABSTRACT: In this article, we study the estimation for a class of semiparametric mixtures of generalized linear models where mixing proportions depend on a covariate non parametrically. We investigate a backfitting estimation procedure and show the asymptotic normality of the proposed estimators under mild conditions. We conduct simulation to show the good performance of our methodology and give a real data analysis as an illustration.