摘要
投影寻踪分类模型作为一种多因素影响问题的综合评价方法,已经被研究者广泛应用在各个领域并取得了良好的效果.然而模型本身还存在密度窗宽不确定以及模型无分类规则等尚需解决的问题.针对这些问题,提出一个基于K-Means动态分类的投影寻踪分类模型,定义了一个新的投影指标.实证分析说明了该模型的可靠性和可操作性.
As a comprehensive evaluation model, projection pursuit classification model has been applied in research widely and gained successful results. However, there are some drawbacks, such as the uncertain cutoff radius, lacking classification rules. In order to solve these problems, a projection pursuit classification model based on K-Means dynamic cluster is proposed and a new projection index is constructed in this paper. Finally, the empirical study shows this model is credible and feasible.
出处
《南京师大学报（自然科学版）》
CAS
CSCD
北大核心
2009年第4期16-20,共5页
Journal of Nanjing Normal University(Natural Science Edition)
基金
国家自然科学基金(60875001)资助项目
关键词
投影寻踪分类
动态聚类
投影指标
遗传算法
projection pursuit classification, dynamic cluster, projection index, genetic algorithm
