It would be great if there was functionality to compute maximum likelihood estimates of distribution parameters from observed data. I'd be happy to work on this - I could directly port the corresponding functionality from the Julia library JuliaStats.jl.