Regression estimation via information-weighted composite models with different dimensions

doi:10.1080/03610918.2019.1586928

CORC > 上海财经大学 > 上海财经大学

	Regression estimation via information-weighted composite models with different dimensions
	Huang, Mian 1; He, Kang 1; Yao, Weixin 2
刊名	COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION
	2019-03-15
关键词	Composite Models EM algorithm Information criteria
ISSN号	0361-0918
DOI	10.1080/03610918.2019.1586928
英文摘要	In this paper, we propose a new class of regression estimation methods by combining many candidate models with possibly different dimensions to address the issue of tapering effect estimation. An information-weighted composite likelihood is proposed. We derive its connection with mixture models and propose an EM algorithm to maximize the composite likelihood. The ascent property of the derived EM algorithm is also established. The simulation study demonstrates the effectiveness of the proposed method.
WOS研究方向	Mathematics
语种	英语
出版者	TAYLOR & FRANCIS INC
WOS记录号	WOS:000465985900001
内容类型	期刊论文
源URL	[http://10.2.47.112/handle/2XS4QKH4/308]
专题	上海财经大学
通讯作者	Yao, Weixin
作者单位	1.Shanghai Univ Finance & Econ, Sch Stat & Management, Shanghai, Peoples R China; 2.Univ Calif Riverside, Dept Stat, Riverside, CA 92521 USA
推荐引用方式 GB/T 7714	Huang, Mian,He, Kang,Yao, Weixin. Regression estimation via information-weighted composite models with different dimensions[J]. COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION,2019.
APA	Huang, Mian,He, Kang,&Yao, Weixin.(2019).Regression estimation via information-weighted composite models with different dimensions.COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION.
MLA	Huang, Mian,et al."Regression estimation via information-weighted composite models with different dimensions".COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION (2019).