摘要
本文依托CSTPC数据库,采用主成分分析法对医学期刊进行评价,然后用评价值作为被解释变量,评价指标作为解释变量进行回归分析,发现部分指标的回归系数为负,删除这几个指标后继续进行主成分分析及回归分析,发现可以删除绝大多数评价指标,得出违背常理的结论。由于学术期刊的特点,不能采取主成分分析进行学术期刊的评价;必须再增加主成分分析回归系数为正这一条件;在评价条件完全具备的情况下,主成分结果回归法(PCARR)可以进行评价指标的筛选;本文提供的方法可广泛应用于某些系统评价方法的检验。
Based on the CSTPC of the Institute of Scientific and Technical Information of China, this paper uses the principal component analysis method to assess medical journals, then makes regression analysis using the evaluation values as the variables to be interpreted and the evaluation indicators as the interpreting variables. The results show some regression coefficients are minus. After deleting these indicators, the paper makes principal component analysis and regression analysis again, and finds out that the great majority of indictors can be deleted, leading to the conclusion contrary to common sense. Due to the characteristics of academic journals, the principal component analysis method can not be used to assess them. The conditions that regression coefficients are plus must be added. Under the circumstances that the evaluation conditions are complete, Principal Component Analysis Result Regression （PCARR） can be used to select indicators. This method can be widely used to examine the evaluation method of some systems.
出处
《情报理论与实践》
CSSCI
北大核心
2009年第9期84-87,共4页
Information Studies:Theory & Application
基金
国家"十一五"支撑计划项目(项目编号:2006BAH-03B05)
国家自然科学基金项目(项目编号:70673019)的研究成果
关键词
学术期刊
评价
主成分分析法
研究
academic journal
evaluation
principal component analysis
research
