bayesian estimation of groups?

[TheChymera]

Can this package do anything along the lines of best?

[jluttine]

I'm not familiar with BEST and didn't get all the details by a quick glance. Anyway, it should be possible to estimate two Gaussian (or Student t) distributions with unknown means and variances (and shared degrees of freedom). For such a simple model, MCMC probably performs very well and is easy to use and all the 1-D plots are easy to make with histograms. With VB, you'll lose some dependencies among the variables. But yes, in principle, I think BayesPy should support something along the lines of BEST, although I believe the results aren't as good as with MCMC because of the ignored posterior dependencies.