Object Orientated BN/MC

[jonititan]

Has anyone used PyMC for object orientated Bayesian networks or object orientated Markov chain Monte Carlo simulations?

I’ve been reading several papers that discuss OOBN and it looks like a nice way to organise things semantically.

Quantitative risk analysis of offshore drilling operations: A Bayesian approach - ScienceDirect

I even found one that used PyMC
Microsoft Word - Saideep_Aviation18_Full.docx (aiaa.org) and this Automated uncertainty quantification analysis using a system model and data | IEEE Conference Publication | IEEE Xplore
They use a graphical editor to create the system model which is then exported to JSON. Their system then reads the JSON and makes an MCMC model out of it in PyMC.

Has anyone else done object orientated models?

[paul_cox]

There was a good discussion back in 2020 on BN and PYMC but I haven’t seen anything since.

Bayes Nets, Belief Networks, and PyMC

So I am trying to get my head around how discrete Bayes Nets (sometimes called Belief Networks) relate to the kind of Bayesian Networks used all the time in PyMC3/STAN/etc. Here’s a concrete example: [bayes_net_1] This can be implemented in pomegranate (just one of the relevant Python packages) as: import pomegranate as pg smokeD = pg.DiscreteDistribution({'yes': 0.25, 'no': 0.75}) covidD = pg.DiscreteDistribution({'yes': 0.1, 'no': 0.9}) hospitalD = pg.ConditionalProbabilityTable( [['y…