摘要
本文讨论判别分析与Logistic回归组合方法的二分类问题.模拟结果显示,基于判别分析的Logistic回归的第一组合法的回判正确率往往高于判别分析分类和Logistic回归分类.第二组合法是基于Logistic回归的判别分析分类,回判正确率接近于Logistic回归的分类结果.很多情况下,这两种方法的回判正确率都高于判别分析。
The combinations of discriminatory analysis and logistic regression could improve to solve binary classification problems. Simulation results show that discriminatory accuracy rate of logistic regression with classification of discriminatory analysis based often more large than ones of discriminatory analysis or logistic regression. Discriminatory accuracy rate of discriminatory analysis with classification of logistic regression approximately equals to one of logistic regression. In many situations, the two methods are often better than discriminatory analysis.
出处
《数理统计与管理》
CSSCI
北大核心
2014年第2期256-265,共10页
Journal of Applied Statistics and Management
基金
教育部新世纪优秀人才支持计划(NCET-08-0909)
教育部留学回国人员科研启动基金项目
全国统计科学研究计划项目(2010LC60
2011LY096)
天津市哲学社会科学规划项目(TJTJ12-004)
天津财经大学科研发展基金项目(Y1204)
关键词
判别分析
LOGISTIC回归
回判正确率
discriminatory analysis, logistic regression, discriminatory accuracy rate
