# Estimation of the parameters of networks from data

**URL:** <https://discourse.pymc.io/t/estimation-of-the-parameters-of-networks-from-data/10331>\
**Category:** Questions\
**Tags:** modeling\
**Created:** [September 8, 2022, 4:54am UTC](https://discourse.pymc.io/t/estimation-of-the-parameters-of-networks-from-data/10331 "2022-09-08T04:54:40Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![xavir](https://yyz2.discourse-cdn.com/flex036/user_avatar/discourse.pymc.io/xavir/32/5598_2.png) [@xavir](https://discourse.pymc.io/u/xavir)\
**Post date:** [September 8, 2022, 4:54am UTC](https://discourse.pymc.io/t/estimation-of-the-parameters-of-networks-from-data/10331/1 "2022-09-08T04:54:40Z")

</div>

Let a dataset _D = {G1, G2, Gi, … Gn}_ where each _Gi_ is a _DAG Gi=(Vi, Ei)_ with node set _Vi_ and edge set _Ei_. Each node in _Vi_ takes a value in set _{red, blue, green}_. The graphs _Gi_ are distinct in the sense that they can have different number of nodes and different edge distributions.

Is there a way to use PyMC to estimate the distribution(s) of the vertices from the dataset _D_? For example, estimate probabilities like _Prob(random node v is red | parent(v) is blue)_?

Thanks.

---

<div class="post-metadata">

**Author:** ![ricardoV94](https://yyz2.discourse-cdn.com/flex036/user_avatar/discourse.pymc.io/ricardov94/32/5775_2.png) [@ricardoV94](https://discourse.pymc.io/u/ricardoV94)\
**Post date:** [September 8, 2022, 5:43am UTC](https://discourse.pymc.io/t/estimation-of-the-parameters-of-networks-from-data/10331/2 "2022-09-08T05:43:43Z")

</div>

Does this discussion help? [Bayes Nets, Belief Networks, and PyMC](https://discourse.pymc.io/t/bayes-nets-belief-networks-and-pymc/5150)

---

<div class="post-metadata">

**Author:** ![xavir](https://yyz2.discourse-cdn.com/flex036/user_avatar/discourse.pymc.io/xavir/32/5598_2.png) [@xavir](https://discourse.pymc.io/u/xavir)\
**Post date:** [September 8, 2022, 6:12am UTC](https://discourse.pymc.io/t/estimation-of-the-parameters-of-networks-from-data/10331/3 "2022-09-08T06:12:11Z")

</div>

Thank you Ricardo for the instructive link. I initially thought about modeling the problem using a Bayesian network but I was not sure about the choice of the random variables, how effective it would be…