摘要
针对模糊时间序列研究中比率划分论域方法存在对非均匀数据划分效果不理想的缺陷,提出一种基于模糊?均值聚类(FCM)算法的多尺度比率划分论域的方法.首先利用FCM算法对样本数据进行分类;然后计算各类数据的平均相对误差,并基于各类的平均误差划分论域,产生非等间隔的多尺度论域划分方法;最后,通过算例表明了多尺度比率论域划分方法的有效性.
In view of the shortcoming of Huarng and Yu＇s ratio method, this paper presents a method of multi-scale ratios interval partitioning. Firstly, the sample data are classified based on the fuzzy C means clustering algorithm. Then, unequal- sized intervals are obtained based on average relative differences of various classes by calculating average relative difference every class. Finally, real examples show the effectiveness of the proposed interval partitioning method.
出处
《控制与决策》
EI
CSCD
北大核心
2014年第2期251-256,共6页
Control and Decision
基金
国家自然科学基金项目(60875032)
中央高校创新团队项目(3132013332)
