Model selection in logistic regression using p-values and greedy search

摘要: We study new logistic model selection criteria based on p-values. The rules are proved to be consistent provided suitable assumptions design matrix and scaling constants satisfied the search is performed over family of all submodels. Moreover, we investigate practical performance introduced in conjunction with greedy methods such as initial ordering, forward backward genetic algorithm which restrict range models an optimal value respective criterion sought. Scaled minimal p-value ordering turns out a promising alternative BIC.