# Problems with model comparison via loo and kfold

**URL:** <https://discourse.mc-stan.org/t/problems-with-model-comparison-via-loo-and-kfold/20613>\
**Category:** Modeling\
**Tags:** loo, brms\
**Created:** [February 4, 2021, 1:38pm UTC](https://discourse.mc-stan.org/t/problems-with-model-comparison-via-loo-and-kfold/20613 "2021-02-04T13:38:43Z")\
**Posts on this page:** 1\
**Showing post:** 3

<div class="post-metadata">

**Author:** ![torkar](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.mc-stan.org/torkar/32/1355_2.png) [@torkar](https://discourse.mc-stan.org/u/torkar)\
**Post date:** [February 7, 2021, 7:49am UTC](https://discourse.mc-stan.org/t/problems-with-model-comparison-via-loo-and-kfold/20613/3 "2021-02-07T07:49:55Z")

</div>

Hi,

even though I believe that you will find that `uvsdt2` will be the better model, relatively speaking, it’s still annoying that you can’t run with `moment_match = TRUE`. Could it be that you are stumbling over the same error as here?

> [@Error with add\_criterion(..., "loo")](https://discourse.mc-stan.org/t/error-with-add-criterion-loo/20475/11):
>
> I think it is probaly a numerical problem during moment matching. I dont know if we have a proper fix for it at this point. You could open an issue on loo github and provide a minimal reproducible example. Perhaps we can eventually find a way around it.

---

_[View the full topic](https://discourse.mc-stan.org/t/problems-with-model-comparison-via-loo-and-kfold/20613)._