bnstruct: an R package for Bayesian Network structure learning in the presence of missing data.

Area of Science:

• Computational statistics
• Machine learning

Background:

• Bayesian Networks are probabilistic graphical models representing dependencies between random variables.
• Learning and inference in Bayesian Networks are crucial for data analysis.
• Handling missing data in these models is a significant practical challenge.

Purpose of the Study:

• Introduce bnstruct, an open-source R package for Bayesian Network analysis.
• Provide a unified tool for structure learning, parameter estimation, and inference.
• Address the common issue of missing data in practical applications.

Main Methods:

• The bnstruct package implements algorithms for Bayesian Network structure and parameter learning.
• It incorporates methods for handling missing data during the learning process.
• The package supports probabilistic reasoning and inference on learned networks.

Main Results:

• bnstruct offers a comprehensive suite of tools for Bayesian Network analysis within R.
• It uniquely handles missing data across structure learning, parameter estimation, and inference tasks.
• The software provides an open-source solution for a common data analysis problem.

Conclusions:

• bnstruct is a valuable open-source tool for researchers and practitioners working with Bayesian Networks.
• Its ability to manage missing data makes it particularly useful for real-world datasets.
• The package facilitates robust analysis and inference in probabilistic graphical models.