bay_gauss = BayesianGaussianMixture(n_components=8, covariance_type='diag', weight_concentration_prior_type='dirichlet_process', weight_concentration_prior=0.1, init_params='kmeans', random_state=12)