I am making a model of earnings by the mixture of normal distributions. I used Cluster analysis in JMP, the option that estimate the parameters by EM algorithm. The output contains counts of statistical units and the proportions of every one cluster. But the counts doesnt reffer to the proportions. In manual is written something about that JMP uses Bayes method. Does it mean that every statistical unit has the probability, that it comes from one of the cluster and the counts depend on the statistical unists whit the highest probability?