基于扩散信息素的蚁群聚类算法及应用

A Pheromone Diffusion based Ant Colony Clustering Algorithm and Applications

本文提出一种基于扩散信息素模型的全局收敛蚁群聚类算法,设计新的信息素更新机制与概率转移机制,适用于复杂的数据集分析。实验结果表明,新算法在聚类效果上比基本的蚁群聚类算法有较明显的改善。最后将新算法应用于电信运营商的客户数据分析中,用于建立客户细分聚类模型,对复杂客户数据集进行分类,取得了较理想的效果。

This paper proposed a pheromone diffusion model based global converging ant colony clustering algorithm（PD-CACCA）,which not only designs the pheromone renewing mechanism but also the probability transferring scheme.PD-CACCA is suitable for analysis of complex data set.Experimental results show that the PD-CACCA algorithm can achieve distinct improvements compared to basic ant colony clustering algorithms.At last,the PD-CACCA is applied to analyze the customer data set for telecom operators,for building customer clustering model which classifies the customers,and the application results are good.