摘要
国家自然科学基金委设立"重点学术期刊专项基金"资助国内优秀学术期刊,在优秀期刊的遴选和资助效果评价过程中,存在期刊数量大,评价指标多等问题.基于此,提出一种针对大规模高维数据的简化的区间数据主成分分析方法(simplified principal component analysis,SPCA).该方法将区间主成分分析分解成2个基本阶段:第一是如何更加简单和高精度地计算数据集合的主轴,第二是如何绘制可视性与可解释性都更强的主平面图,以增强研究人员对大规模数据主要特征的洞察能力.采用SPCA方法,一方面能够从整体上研究各学科期刊的差异,说明学科之间的不可比较性;另一方面能够筛选出衡量期刊水平的关键指标,遴选优秀期刊,同时分析连续受资助期刊的动态资助效果.
National Natural Science Foundation of China （NSFC） encountered many problems, such as an increasing number of applicants and a great many indices, when selecting excellent academic journals as grantees of ＂Key Academic Journal Fund＂ and evaluating funding effects. This paper puts forward a method for Principal Component Analysis on interval data, referred to as Simplified Principal Component Analysis （SPCA） , specific for large-scaled and high-dimensional data. The proposed method involves two steps, first look for factor axes of the original large-scaled data and then illustrate the analytical results on a low-dimensional space with strengthened visibility and high interpretation. Performing SPCA on academic journal dataset helps to understand the differences in all disciplines and to choose key criteria for selecting excellent journals and measuring funding effect dynamically.
出处
《管理科学学报》
CSSCI
北大核心
2010年第7期88-94,共7页
Journal of Management Sciences in China
基金
国家自然科学基金创新研究群体科学基金资助项目(70821061)
国家自然科学基金重点资助项目(70531010)
国家自然科学基金资助项目(70771004
J0824301)
关键词
区间数据
SPCA
主平面图
期刊评价
资助效果分析
interval data
SPCA
the plane of principal axes
journal evaluation
funding effect analysis
