depynd - evaluating dependencies among random variables

depynd is a Python library for evaluating dependencies among random variables from data. It supports learning statistical dependencies for one-to-one, one-to-many, and many-to-many relationships, where each one corresponds to

• depynd.information: mutual information (MI) estimation,
• depynd.feature_selection: feature selection, and
• depynd.markov_networks: Markov network structure learning,

respectively. Specifically, depynd supports MI estimation for discrete-continuous mixtures, MI-based feature selection, and structure learning of Markov networks (a.k.a. Markov random fields).