蚁群聚类算法研究与应用.pdfVIP

摘 要 数据挖掘是从海量的数据中寻找模式或规则的过程。聚类分析作为数据挖掘中一个重 要的研究内容，既可以作为独立的数据挖掘工具，发现数据库中数据分布的一些深入信息， 也可以作为其它数据挖掘算法的预处理步骤，已经被广泛应用在诸多领域。蚁群算法是一 种模拟蚂蚁群体行为的仿生优化算法，因其具有分布式计算、信息正反馈和启发式有哪些信誉好的足球投注网站的 特征在聚类分析中显示出很好的应用前景。 本文在充分研究了现有蚁群聚类算法的基本原理与特性的基础上，为了进一步提高算 法的收敛速度，改善聚类质量，提出了基于蚁堆形成原理的改进的蚁群聚类算法，并将该 算法与一种改进的遗传算法进行融合。主要思想是利用改进的遗传算法的快速全局有哪些信誉好的足球投注网站能 力，生成蚁群聚类算法中的信息素初始分布，然后再利用蚁群聚类算法的正反馈机制求精 确解。该算法实现了两个算法的优势互补，大大减低了蚁群聚类算法中对参数的调整而造 成的试验次数过多，并有效地防止了蚁群聚类算法很快地收敛到局部最优解的缺点，使得 算法在优化性能和时间性能两个方面都有所改善。最后将该算法应用于房地产行业的客户 细分中。 关键词：蚁群聚类算法；遗传算法；聚类分析；客户细分；房地产行业 - I - Abstract Data mining is the process of automatically searching large volumes of data for patterns and rules. Cluster analysis, as one of the important research branch of data mining, has been widely applied to many other domains. It can be used as the tool of data mining to discover the in-deep information of data distribution in database, and also be considered as the pretreatment process of other data mining algorithms. As a new intelligent bionic algorithm, ant colony algorithm has showed good prospect in the cluster analysis because it has features of distributed parallel caleulation, information positive feedback and heuristic search ability. Based on fully studying the existing basic principle and characteristics of ant clustering algorithm. This paper proposed improved ant clustering algorithm which is based on ant pile of forming principle and fusion with an improved genetic algorithm, in order to further improve the algorithm convergence speed and cluster quality. Main idea of this algorithm is firstly generating initial distribution of pheromone for ant clustering algorithm using the rapid global search capability of the improved genetic algorithm. Then get the exact solution using positive feedback of ant clustering algorithm. This a