A novel cluster analysis method supervised by statistical tests is proposed in this paper, which processes three key problems in data analysis, cluster tendency, cluster analysis and cluster validity, simultaneously. So, it provides an analysis tool for the validity and reasonableness of pattern unsupervised classification, especially in the case of large number of samples. The experimental results demonstrate its effectiveness.
Anderberg M R.Cluster Analysis for Applications.New York:Academic Press,1973:chapter 1[2]Bezdek J C.Pattern Recognition with Fhzzy Objective Function Algorithms.NewYork:Plenum Pyess,1981.[3]Panayirci E,Dubes R C.A test for multidimensional clustering tendency.Pattern Reconition,Panayirci E,Dubes R C.A test for multidimensional clustering tendency.Pattern Reconition,Panayirci E,1983,16(4):433-444.[4]Dubes R C:Jain A K.Validity studies in clustering methodologies.Pattern Recognition, 1979,12(11):235-247.[5]Besag J E、Gleaves J T On the detection of spatial pattern in plant communities.Bulletin of the International StatisticaI Institute,1973:45(1):153-158.[6]曾广周,随机分布模式的统计检验.山东工业大学学报,1986,16(4):12-18.[7]高新波.基于进化计算和神经网络的模糊聚类新算法研究. [硕士论文].西安:西安电子科技大学,1996.[8]Gao Xinbo,et al. An initialization method for multi-type prototype fuzzy clustering. Proceeding of International Conference on Signal Processing:ICSP98,Beeijing:1998,1205-1208.[9]裴继红.基于模糊信息处理的图象分割方法研究:[博士论文].西安:西安电子科技大学,1998.[10]Hoffman R,Jain A K.A test of randomness besed on the minimal spanning tree.Pattern Recognition Lett.,1983, 4(1):175-180.