Mixed Model-based Cluster Methods

The cluster analysis is a branch of multivariate statistics.Mixed model-based cluster algorithm is one of the numerous cluster algorithms.In this paper, I comprehensively discuss the general the-ory frame of mixed model-based cluster algorithm,and cluster algorithms,I develope a new cluster algorithm, joint finite mixed mixture model for clustering from independent Gaussian and beta dis-tributed data,which called BGMMn. By the simulation,giving the comparison between the cluster algorithm of BGMMn.BGMMs,BGMMa,BGMMh,on the clustering accuracy and the correct number of clusters.Thus,point out that the BGMMn cluster algorithm is better than the other three cluster algorithms,and the BGMMn cluster algorithm is better for useing, when Gaussian distributed datas are easy to cluster.