I also have another question. So I am running a bernoulli mixture model using your package and I noticed that I am getting different size clusters when I run multiple iterations using the same size K for the update() . I don't know the true number of clusters for my dataset, but the cluster sizes also fluctuate as a vary the upper bound on K. Any insight as to why the number of clusters are varying? My first thought of why the cluster sizes are varying is due to the lack of similarity in the dataset. The dimension of my dataset is (3448, 174) of binary values.