Posterior distribution clustering

My first thought would be to apply k-means to your posterior samples. Do you need some sort of downstream uncertainty quantification or other issue that would require using a Bayesian model for it?

1 Like