科研综合实力评价方法研究

Research on Evaluation Method of Comprehensive Strength

摘要: 本文收集2013年31个省市自治区部分高校有关人文社科科研方面的相关数据，利用多元统计方法对科研综合实力进行评价研究。针对主成分综合评价存在的问题，提出一些改进措施，给出了主成分综合评价的应用条件；以信息熵为工具，利用指标的变异程度度量信息差异，为科研综合实力评价提供方法参考；若第一主成分的评价结果与熵值法的评价结果通过Kendall协同系数检验，可以集成新的评价方案；通过对主成分得分进行k-均值聚类，并以最终聚类中心为类间排序标准，以第一主成分评分作为类内评价依据综合考量科研综合实力。

Abstract: In this paper, we collect the data about the social science research of the universities throughout 31 provinces during a year, and then evaluate the comprehensive strength of scientific research by the multivariate statistical methods. In order to improve the comprehensive evaluation of the principal components, we introduce the application conditions. With the information entropy as a tool, we use the information difference of the index of the degree of variation of indicators to provide the basis for the comprehensive strength evaluation of scientific research. If the first principal component evaluation results and entropy method evaluation results pass the Kendall coordination coefficient test, then we can integrate the above two solutions to form a final evaluation results. We evaluate the comprehensive strength of scientific research by k-means clustering for principal component scores, choosing the final cluster center as the inter class sort criteria, and taking the first principal component score as the criteria of the inner class evaluation.