摘要
对自然联系函数下广义线性模型的变量选择问题进行了研究.首先,分别按Ward准则和似然比准则,给出了两种变量选择的方法;然后,在一定的条件下,证明了上述两种变量选择方法的弱相合性.
The variable selection in generalized linear model under the natural link function was studied. First, corresponding to Wald Criterion and Likelihood Criterion, two selection procedures were presented. Then, under certain conditions, it was proved that the above selection procedures are of weak consistency.
基金
国家自然科学基金(1017109410571001)
教育部博士点基金(20030358053)
中国科学院知识创新工程资助
南京师范大学科研启动基金(2005101XGQ2B84)资助
关键词
广义线性模型
WALD检验
似然比检验
变量选择
generalized linear model
Wald test
likelihood ration test
variable selection
