MIxBN: library for learning Bayesian networks from mixed data

TL;DR

This paper introduces MIxBN, a library for learning Bayesian networks from mixed data containing discrete and continuous variables, using novel algorithms that avoid data discretization to preserve information.

Contribution

It presents a new algorithm for structural and parameter learning from mixed data without discretization, and provides two algorithms for graph structure enumeration, enhancing Bayesian network learning.

Findings

Effective learning from mixed data demonstrated on synthetic datasets.

Improved accuracy over discretization-based methods.

Versatile algorithms for structure enumeration tested on real datasets.

Abstract

This paper describes a new library for learning Bayesian networks from data containing discrete and continuous variables (mixed data). In addition to the classical learning methods on discretized data, this library proposes its algorithm that allows structural learning and parameters learning from mixed data without discretization since data discretization leads to information loss. This algorithm based on mixed MI score function for structural learning, and also linear regression and Gaussian distribution approximation for parameters learning. The library also offers two algorithms for enumerating graph structures - the greedy Hill-Climbing algorithm and the evolutionary algorithm. Thus the key capabilities of the proposed library are as follows: (1) structural and parameters learning of a Bayesian network on discretized data, (2) structural and parameters learning of a Bayesian network on mixed data using the MI mixed score function and Gaussian approximation, (3) launching learning algorithms on one of two algorithms for enumerating graph structures - Hill-Climbing and the evolutionary algorithm. Since the need for mixed data representation comes from practical necessity, the advantages of our implementations are evaluated in the context of solving approximation and gap recovery problems on synthetic data and real datasets.