Liu Xiaoqian, Lecturer of School of Economics and Finance (SEF), has recently published her paper “A modified multinomial baseline logit model with logit functions having different covariates” (corresponding author) in the journal Communications in Statistics - Simulation and Computation (Volume 49 Issue 11:2861-2875,DOI:https://doi.org/10.1080/03610918.2018.1529238).
The paper presents a modified multinomial baseline logit model and gives the estimation procedure for the model. In the traditional multinomial logistic regression, same covariates in the logistic function are required in baseline logit regression, which causes inconvenience in the analysis of some polychotomous data. For example, some covariates may not exist in one logit function. The improved baseline logit model proposed in this paper can have different covariates. The research results contribute to the study of multivariate logistic regression methods in generalized linear models, providing an alternative data analysis method for multivariate logistic regression and enriching and developing the relevant theoretical research on the research problem of multivariate logit regression model. Compared with existing methods of variable selection, the method outperforms for an appropriate number of covariates, and the model has potential applications in disciplines such as finance, sociology, industry, and agriculture.
Since joining SEF, Dr. Liu Xiaoqian has published more than 10 papers in authoritative journals, including some indexed in SCI and EI. She has completed one youth program of NSFC and one MOE (Ministry of Education) youth program of humanities and social sciences. She also guided graduates and undergraduates to win national awards in the Mathematical Contest in Modeling for many times.