摘要
面板数据由于能够从截面和时间构成的二维空间来描述研究对象的动态特征而被广泛应用于经济问题的建模实践中。本文借鉴多元统计学中主成分分析方法对面板数据进行降维处理,然后通过构建综合评价函数序列矩阵的相似指标对面板数据进行聚类分析,并提出一些研究面板数据亲疏关系的有效途径,最后运用该算法对我国地区科技能力进行实证分析,结果与实际状况较为吻合.
Panel data is widely used in modeling on economic problems because it can describe the object＇s dynamic characters from time and cross-section in two-dimensional space. This paper reduces dimension of panel data by the means of principal component analysis, then uses cluster analysis on the panel data through similarity indexes which describe the comprehensive sequential evaluation matrix, lastly puts forward the effective research approaches on the panel data＇ s affinities. After that the authors use this algorithm to analyze the regional capacity of science and technology in China, analysis result comparatively tallies with actual situation.
出处
《数理统计与管理》
CSSCI
北大核心
2009年第5期831-838,共8页
Journal of Applied Statistics and Management
基金
国家自然科学基金项目(704730370)
湖北省统计局重点项目(hb091-18)
武汉市社科基金项目(08022)
关键词
面板数据
聚类分析
主成分分析
panel data, cluster analysis, principal component analysis
