摘要
基于大学生对于COVID-19认知数据以及影响因素,建立了多类别logistic回归模型。对比Group Lasso惩罚对数似然和逐步回归两种变量选择方法,比较AIC、BIC准则和交叉验证三种模型选择方法的模型效果,得到最优模型来分析影响学生对COVID-19相关知识掌握情况。
Based on student COVID-19 cognitive data and influential factors,we build a multi-class logistic regression model.Variable selection methods such as Group Lasso penalized log likelihood and stepwise regression are compared,and model selection methods such as AIC criterion,BIC criterion and cross validation are discussed to optimize the model for analyzing the factors influencing student cognition to COVID-19.
作者
段萱健
徐平峰
DUAN Xuanjian;XU Pingfeng(School of Mathematics and Statistics, Changchun University of Technology, Changchun 130012, China)
出处
《长春工业大学学报》
CAS
2022年第1期53-57,共5页
Journal of Changchun University of Technology
基金
吉林省自然科学基金资助项目(20210101152JC)。
