【摘要】 结合动态聚类思想,对投影寻踪聚类模型进行改进,建立了投影寻踪动态聚类模型.针对多因素聚类问题的高维复杂性,利用线性投影技术将其转换为关于投影特征值的线性聚类问题;根据动态聚类思想构建新的投影指标,对投影特征值序列进行动态聚类,进而在低维空间实现高维数据样本的聚类分析;实证分析验证了投影寻踪动态聚类模型的可行性与有效性.结果表明,投影寻踪动态聚类模型实现了投影寻踪与动态聚类理论的有机结合,操作简便,无人为干扰,为多因素样本聚类分析提供了一种新方法.更多还原

【Abstract】 The present study develops a projection pursuit dynamic cluster(PPDC) model by combining dynamic cluster with projection pursuit to solve the problem of water security evaluation for some provinces of China.Firstly,a multi-factor cluster problem was converted into a single-factor(projected characteristic value) cluster problem according to linear projection.Secondly,a new projection index on the basis of dynamic cluster rule was set up in the PPDC model,which successfully avoids the problem of parameter calibration and makes cluster results objective.Finally,the developed PPDC model was used in a case study of water security evaluation for some provinces of China to evaluate its application.Cluster results of the PPDC model agree well with those of the fuzzy optimization,indicating that the PPDC model is a powerful tool in multi-factor cluster analyses and can become be a new method for water security evaluation.更多还原