Python Scikit Find out GMM results incompatible with R Mclust
I am trying to reproduce the results in Scikit Learn GMM of MClust in R. With the data, the results I get are different across all packages. I have tried different covariance structures in the mixture. Hmm. How do I get the Python version? The simpler examples work fine, but with this variance structure, I cannot get it to work.
Python code:
gmm = mixture.GMM(n_components=3,n_iter=1000,covariance_type='full')
gmm.fit(data)
gmm.means_
array([[ 0.08603919],
[ 0.08590469],
[ 0.08617066]])
gmm.covars_
array([[ 0.00122368],
[ 0.0012216 ],
[ 0.00122569]])
R-Code
res<-Mclust(Stamp$thickness)
res$param$mean
0.07215458 0.07935341 0.09919740
res$param$variance$sigmasq
4.814927e-06 3.097694e-06 1.884615e-04
+3
source to share