基于logistic模型的属性数据的研究和应用  

Research and Application of Attribute Data Based on Logistic Model

作　　者：王佳璇 孙聚波 WANG Jia-xuan;SUN Ju-bo(Jilin Engineering Normal University, Changchun Jilin 130052, China)

机构地区：[1]吉林工程技术师范学院,吉林长春130052

出　　处：《吉林工程技术师范学院学报》2021年第3期88-92,共5页Journal of Jilin Engineering Normal University

摘　　要：属性响应变量的分布是非正态的,且与解释变量的关系是非线性的,常用广义线性模型分析处理它。广义线性模型中最为流行的是logistic模型。首先研究了Logistic模型在二分类数据分析中的性质,并分析了近年来某电商网站上顾客购买国内外各品牌维生素的数据。文中还研究了Logistic模型在多分类情形下的拓展,结合实例分析了名义响应变量的基线-类别logit模型和有序响应变量的累积logit模型。The distribution of an attribute responsive variable is non-normal,and its relationship with the explanatory variable is nonlinear.Generally,its analysis depends on the generalized linear models,of which the most popular one is the logistic model.The paper first studies the nature of the Logistic model in binary data analysis,and analyzes the data on the purchase of various domestic and foreign brands of vitamins by customers on an e-commerce website in recent years.Additionally,it also studies the expansion of the logistic model in the multi-category situation,and analyzes the baseline-category logit model of the nominal responsive variables and the cumulative logit model of the ordered responsive variables on the basis of some instances.

关 键 词：属性数据 LOGISTIC模型 基线-类别logit模型 累积logit模型 

分 类 号：O212[理学—概率论与数理统计]