Fork me on GitHub
#data-science
<
2017-07-23
>
nblumoe05:07:53

If performance is a concern and you can accept the restrictions, I would try Bayadera tbh

nblumoe05:07:10

Also because you already have your models and just want/nees a faster sampler as it seems.

nblumoe05:07:32

What are you using so far? Own, custom code?

nblumoe05:07:08

I have not used Bayadera myself, neither am I aware of a “proper” documentation. There are examples for the Kruschke book in bayadera though, which is cool: https://github.com/uncomplicate/bayadera/tree/master/test/clojure/uncomplicate/bayadera/examples/dbda

nblumoe05:07:48

Aaaaand, what to use also depends on what kind of MCMC you actually want to do and if that is supported. 🙂

elise_huard07:07:54

@nblumoe: custom code 🙂

elise_huard07:07:30

will have a look, thanks