文章摘要
张弘,赵惠祥,刘燕萍,陶文文.基于主成分分析法的科技期刊评价方法.编辑学报,2008,20(1):87-90
基于主成分分析法的科技期刊评价方法
Evaluation method on sci-tech journals based principle component anal ysis
投稿时间:2007-09-02  修订日期:2007-09-02
DOI:
中文关键词: 科技期刊  期刊评价  主成分分析  相关系数矩阵  特征 向量
英文关键词: sci-tech journal  journal evaluation  principle comp onent analysis  correlation coefficient matrix  eigenvector
基金项目:
作者单位
张弘 同济大学学报自然科学版编辑部,200092,上海 
赵惠祥 同济大学学报自然科学版编辑部,200092,上海 
刘燕萍 同济大学学报自然科学版编辑部,200092,上海 
陶文文 同济大学学报自然科学版编辑部,200092,上海 
摘要点击次数: 775
全文下载次数: 387
中文摘要:
      针对目前科技期刊评价方法中存在的指标之间相关性问题和 指标权重选取问题,提出了基于主成分分析的科技期刊评价方法。该评价方法通过相关系数 矩阵的特征向量将评价指标线性变化成彼此独立的主成分,根据主成分累计贡献值确定主成 分的取用维数,由主成分方差确定权重。其优点是:可以消除由于指标间的相关性带来的偏 差,降低计算维数,从而降低指标选择的难度,提高评价结果的可信度;此外,可以消除人 为确定指标权重引起的弊病,使评价结果更具客观性和准确性。
英文摘要:
      An evaluation method on sci-tech journals based on principal component analysis is proposed to deal with the relevance between various indices and the selection of index weight. According to this evaluation method, the evaluation index is turned into mutually independent principal components on the basis of linear transformation through the eigenvector of correlation coefficient matrix. The dimension selection of the principal components is determined according to the accumulated contributing value of the principal components and the weight is determined by the variance of the principal components. When this method is employed to evaluate the sci-tech journals, the deviation brought about by the relevance between various indices can be eliminated and the calculating dimension can be lowered, hence making the selection of index easier and improving the credibility of the evaluation results. In addition, the problems arising from subjective determination of index weight can be solved, and the evaluation results tend to be more objective and accurate.
查看全文   查看/发表评论  下载PDF阅读器
关闭