Recently open sourced Python for library for Bayesian belief networks (aka Bayesian networks, aka probabilistic graphical models).<p>These are a powerful tool for representing dependence relationships in probability distributions. Given a joint probability distribution, Pr(X1, X2, ..., Xk), a table representing this requires |X1|X...X|Xk| entries. This representation can be accomplished much more compactly by identifying conditional independence relationships among the variables. Belief networks, are one method of encoding these independence relationships. Once this network is identified, it can then be used to perform various types of probabilistic inference.