期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2013
卷号：55
期号：1
出版社：Journal of Theoretical and Applied
摘要：In this paper, we propose a new learning method for clustering heterogeneous data with continuous class. This method in a first step finds the optimal path between the data using ant colony algorithms. The distance adopted in our optimization method takes into account all types of data. In the second step, instances in the optimal path, are divided into homogeneous groups. A new criterion for the separation of clusters is used; it is based on transition probabilities between the instances. A third step is to find the prototype of each cluster to identify the cluster membership of any new data injected. After applying a clustering algorithm, we want to know whether the cluster structure found is valid or not. To validate our approach, we have applied our method on different types of artificial data and real data from UCI Machine Learning Repository. The results obtained showed an obvious improvement of validation indexes compared to those of ACA, ACOC, k-means and KHM algorithms.
关键词：Data clustering; Best path; Cluster prototype; Continuous class.
