Bayesian Parameter estimation shouldn't add Discrete CPDs

pgmpy / pgmpy

Python Library for learning (Structure and Parameter), inference (Probabilistic and Causal), and simulations in Bayesian Networks.

https://pgmpy.org/

MIT License

2.65k stars 696 forks source link

Bayesian Parameter estimation shouldn't add Discrete CPDs #768

Open ankurankan opened 7 years ago

ankurankan commented 7 years ago

Currently Bayesian parameter estimation adds TabularCPD to Bayesian Networks having the MAP values. But it should return a ContinuousCPD object.

yashu-seth commented 7 years ago

@ankurankan Currently we do not support ContinuousCPD object. And what about the discrete cases? Are we planning to have a common class for both continuous and discrete cases as discussed earlier?

ankurankan commented 7 years ago

@yashu-seth I am thinking of having separate classes for discrete and continuous. But I am still thinking about how to implement it.