分类树中QUEST算法与多水平logistic模型的联合应用与比较

目的探讨分类树中的QUEST算法和多水平logistic模型联合应用于分析层次资料的可行性.方法首先介绍模型的原理,据此提出联合应用的具体思路,然后采用分析实例进行深入探讨,并用预测准确率和ROC曲线对模型拟合效果加以比较.结果QUEST算法能非常准确地搜索出主要影响因素,并能更加深入地刻画因变量、自变量间的复杂联系,在此基础上再进行多水平logistic模型会更加高效、准确.结论QUEST算法和多水平logistic模型各有优势,将它们联合用于层次资料的分析将更能保证分析的正确性和完善性.

Application of Combining QUEST Arithmetic in Classification Tree with Multilevel Logistic Model in Analyzing a Real Data Concerning Prenatal Care

Objective To explore the possibility of combining application of QUEST arithmetic with Multilevel logistic model.Methods Based on its principle,a real data was used to examine the feasibility of the method.The approaches of ROC curve and prediction probability were applied in the analysis.Results The complex relationship between dependent and independent variables were accurately found by QUEST arithmetic.To be used together with QUEST arithmetic , the multilevel logistic model would be more efficient and accurate.Conclusion Both QUEST arithmetic and multilevel logistic model had their own advantages and the accuracy of analyzing the results could be guaranteed by combining application of two methods while used in multilevel data.