科技评价中因子分析信息损失的改进

Improvement of Factor Analysis Information Loss in Science and Technology Evaluation

在科技评价中,因子分析是非常重要的评价方法之一,但是存在着牺牲原始数据信息的缺陷。文章提出了采用全部公共因子评价的完全信息因子分析法,以及采用可解释公共因子的最大信息因子分析方法。基于JCR2015经济学期刊的研究表明:因子分析法评价会损失原始数据的大量信息;完全信息因子分析评价弥补了因子分析评价的不足;因子分析评价与完全信息因子分析评价相比虽然相关度较高但排序相差较大;最大信息因子分析最大限度保留了原始数据中的信息,评价结果与完全信息因子分析高度相关并且基本一致,而且方便主观赋权,克服了完全信息因子分析完全进行客观赋权的不足,可根据需要灵活选择完全信息因子分析方法或最大信息因子分析方法进行评价;在使用传统因子分析时应该重分析轻评价,以分析为主。

Factor analysis is one of the most important evaluation methods in science and technology evaluation,but it has a flaw of sacrificing raw data information.This paper proposes the all information factor analysis method and using the maximum information factor analysis method that can explain the common factors.The study based on the JCR2015 Economics Journal shows that factor analysis may lose a lot of information of the original data;complete information factor analysis makes up the deficiency of factor analysis;compared with the complete information factor analysis evaluation,although evaluation of factor analysis has higher correlation,its sorting is quite different;maximum information factor analysis maximizes the retention of information in the original data,by use of which the evaluation results are highly related to and basically consistent with the complete information factor analysis,and convenient for subjective weighting,overcoming the shortage from completely objective weighting in the analysis of complete information factor,able to flexibly select the full information factor analysis method or the maximum information factor analysis method according to the needs;when using traditional factor analysis,we should place more emphasis on analysis instead of evaluation.