摘要
[目的/意义]根据误差理论,测量误差分为系统误差与偶然误差,将其推广到主成分与因子分析,研究误差对指标体系学术评价的影响具有重要意义。[过程/方法]在理论分析的基础上,对主成分与因子分析系统误差的形成机制进行深入分析,探讨检验与测度方法,并以中国知网环境科学与技术期刊为例进行说明。[结果/结论]主成分分析与因子分析普遍存在系统误差;主成分分析与因子分析系统误差产生的环节是多样的,包括评价指标选取、反向指标标准化方法、主成分或公共因子解释力的确定、选取少数主成分或公共因子导致的数据遗失、权重设定、评价值公布方法等;主成分与因子分析的系统误差只可以适当测度;应采取全面措施降低主成分与因子分析的系统误差。
[Purpose/significance]According to the error theory,measurement error is divided into systematic error and accidental error,which is extended to principal component and factor analysis,and it is of great significance to study the influence of error on the academic evaluation of the index system.[Method/process]On the basis of theoretical analysis,the formation mechanism of principal component and factor analysis system error is deeply analyzed,and the test and measurement methods are discussed.[Result/conclusion]Taking CNKI environmental science and technology journals as an example,the results show that there are systematic errors in principal component analysis and factor analysis.The links that generate systematic errors in principal component analysis and factor analysis are diverse,including the selection of evaluation indicators,the standardization method of reverse indicators,the determination of the explanatory power of principal components or public factors,the data loss caused by the selection of a few principal components or public factors,the weight setting,the evaluation value publishing method,etc.The systematic error of principal component analysis and factor analysis can only be measured appropriately.Comprehensive measures should be taken to reduce the systematic error of principal component analysis and factor analysis.
作者
黄龙卫
俞立平
Huang Longwei;Yu Liping(School of Marxism Studies,Shanxi University,Shanxi Taiyuan 030006;Collaborative Innovation Center of Statistical Data Engineering Technology&Application,Zhejiang Gongshang University,Zhejiang Hangzhou 310018)
出处
《情报理论与实践》
北大核心
2025年第4期72-79,71,共9页
Information Studies:Theory & Application
基金
国家社会科学基金教育学一般项目“浸润教育家精神的乡村教师教育课程一体化建设研究”的成果,项目编号:BRA240226。
关键词
学术期刊
学术评价
系统误差
主成分分析
因子分析
academic journal
academic evaluation
system error
principal component analysis
factor analysis
