# Parameter inference for agents based model

**URL:** <https://discourse.julialang.org/t/parameter-inference-for-agents-based-model/118613>\
**Category:** Modelling & Simulations\
**Tags:** turing, agents, eventqueueabm\
**Created:** [August 26, 2024, 8:50am UTC](https://discourse.julialang.org/t/parameter-inference-for-agents-based-model/118613 "2024-08-26T08:50:08Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![Sahil\_Khan](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/sahil_khan/32/47573_2.png) [@Sahil\_Khan](https://discourse.julialang.org/u/Sahil_Khan)\
**Post date:** [August 26, 2024, 8:50am UTC](https://discourse.julialang.org/t/parameter-inference-for-agents-based-model/118613/1 "2024-08-26T08:50:08Z")

</div>

Hello everyone, Im doing parameter inference for my agent based model. At the moment, Im using NUTS() from Turing package. Im not sure if its the right choice as model is stochastic in nature. I was wondering that if a ABC method is much more better ?

Do you guys have suggestions?

---

<div class="post-metadata">

**Author:** ![SteffenPL](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/steffenpl/32/206270_2.png) [@SteffenPL](https://discourse.julialang.org/u/SteffenPL)\
**Post date:** [August 26, 2024, 8:54am UTC](https://discourse.julialang.org/t/parameter-inference-for-agents-based-model/118613/2 "2024-08-26T08:54:42Z")

</div>

What kind of agent-based model and what kind of data do you have? That can make a big difference in what methods to use…

---

<div class="post-metadata">

**Author:** ![Sahil\_Khan](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/sahil_khan/32/47573_2.png) [@Sahil\_Khan](https://discourse.julialang.org/u/Sahil_Khan)\
**Post date:** [August 26, 2024, 8:59am UTC](https://discourse.julialang.org/t/parameter-inference-for-agents-based-model/118613/3 "2024-08-26T08:59:39Z")

</div>

I have a continuous time agent based model. So, in terms of `Agents.jl --- EventBasedModel`. I have some experimental data(dose-response) that Im trying to fit with my model(which try to replicate experimental process). Now model output is not deterministic in nature which is leading to issues during inference with Turing.

---

<div class="post-metadata">

**Author:** ![sdwfrost](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/sdwfrost/32/2831_2.png) [@sdwfrost](https://discourse.julialang.org/u/sdwfrost)\
**Post date:** [August 26, 2024, 11:41pm UTC](https://discourse.julialang.org/t/parameter-inference-for-agents-based-model/118613/4 "2024-08-26T23:41:39Z")

</div>

Have you tried using Gen.jl instead? The tooling lends itself much better to your scenario (though you may have to implement a custom sampler).