# WAIC, or LOO etc in Turing?

**URL:** https://discourse.julialang.org/t/waic-or-loo-etc-in-turing/30873
**Category:** Probabilistic Programming
**Created:** [November 8, 2019, 8:48pm UTC](https://discourse.julialang.org/t/waic-or-loo-etc-in-turing/30873 "2019-11-08T20:48:55Z")
**Posts on this page:** 1
**Showing post:** 8

<div class="post-metadata">

### Author: ![sethaxen](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/sethaxen/32/35604_2.png) [@sethaxen](https://discourse.julialang.org/u/sethaxen)
#### Post date: [November 15, 2019, 7:50pm UTC](https://discourse.julialang.org/t/waic-or-loo-etc-in-turing/30873/8 "2019-11-15T19:50:35Z")

</div>

You’re quite welcome!

`convert_to_inference_data` is a pretty feature-sparse conversion of a data type to an `InferenceData`. In this case, it takes the data in `stan_chns` and converts it. Stan’s output doesn’t differentiate between generated quantities and parameters and moreover between log likelihood and posterior predictions. If you want generated quantities to be treated specially in ArviZ, you’ll want to use a specialized conversion function.

In ArviZ, these start with `from_`. `from_mcmcchains` or `from_cmdstan` will work for you. Check out the [worked example using CmdStan on 8 schools](https://arviz-devs.github.io/ArviZ.jl/dev/quickstart/#Plotting-with-CmdStan.jl-outputs-1) in the quickstart for the signature.

---

_[View the full topic](https://discourse.julialang.org/t/waic-or-loo-etc-in-turing/30873)._