摘要
目的 :探讨模型对未知样本的预测能力的优劣。方法 :介绍相关主成分回归 CPCR法并将该法与传统的主成分回归进行比较。结果 :当所含主成分数较小时 ,CPCR法优于传统的主成分回归。结论 :该方法也是一种可行的多元回归方法。
Objective:To explore the predicting ability of model to unknown samples.Methods:To introduce the correlation principal components regression(CPCR) and compare the difference between the CPCR and top-down PCR.Results:When there are less principal components,CPCR is better than top-down PCR.Conclusion:With improvement of theory,algebra and relevant software,CPCR is a practicable multivariate regression method.
出处
《现代预防医学》
CAS
2002年第3期337-338,共2页
Modern Preventive Medicine
