# Justification for the use of metamodels

**URL:** <https://uqworld.org/t/justification-for-the-use-of-metamodels/1542>\
**Category:** UQ Discussion Forum\
**Tags:** metamodeling\
**Created:** [March 28, 2022, 10:50pm UTC](https://uqworld.org/t/justification-for-the-use-of-metamodels/1542 "2022-03-28T22:50:40Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![HZhang](https://uqworld.org/letter_avatar_proxy/v4/letter/h/df788c/32.png) [@HZhang](https://uqworld.org/u/HZhang)\
**Post date:** [March 28, 2022, 10:50pm UTC](https://uqworld.org/t/justification-for-the-use-of-metamodels/1542/1 "2022-03-28T22:50:41Z")

</div>

Hi everyone,

When we use metamodels in UQ, the metamodels themselves can introduce errors/uncertainties in the results. How do we justify the use of metamodels? Is there any papers specifically on this issue?

---

<div class="post-metadata">

**Author:** ![TamadurAB](https://uqworld.org/letter_avatar_proxy/v4/letter/t/c4cdca/32.png) [@TamadurAB](https://uqworld.org/u/TamadurAB)\
**Post date:** [July 22, 2022, 9:13am UTC](https://uqworld.org/t/justification-for-the-use-of-metamodels/1542/2 "2022-07-22T09:13:46Z")

</div>

Hey Hongzhou,  
Maybe you can check: [https://www.researchgate.net/publication/317396793\_Surrogate\_models\_for\_uncertainty\_quantification\_An\_overview](https://www.researchgate.net/publication/317396793_Surrogate_models_for_uncertainty_quantification_An_overview)

From my point of view, we validate the surrogate model and quantify the errors within it, this is why for example we measure LOO error, validation error and modified validation errors.  
Hope this would help you?  
Regards

---

<div class="post-metadata">

**Author:** ![HZhang](https://uqworld.org/letter_avatar_proxy/v4/letter/h/df788c/32.png) [@HZhang](https://uqworld.org/u/HZhang)\
**Post date:** [October 3, 2022, 9:30pm UTC](https://uqworld.org/t/justification-for-the-use-of-metamodels/1542/3 "2022-10-03T21:30:35Z")

</div>

Thank you very much, Tamadur, for the reference. And yes you are right, having a low validation error of the surrogate model is a good justification of implementing it.

Best,

Hongzhou