Explainable Bayesian networks in R

Link to coding resources for the thesis titled Explainable Bayesian networks: taxonomy, properties and approximation methods. The link provides access to the full repository containing the code and resources for all experiments conducted. The files are as follows:

• Chapter_3: experiments conducted in chapter 3 are based on the insurance Bayesian network (a built-in, benchmark data set) available through the bnlearn package in R. Experiments include a brute-force MAP/MPE, brute-force MRE, and forward-search algorithm for explanation of evidence and same-decision probability implementation for explanation of decisions.
• Chapter_4: experiments conducted in chapter 4 and is based on several benchmark Bayesian networks available through the bnlearn package in R. This includes experiments on the proposed forward-gLasso search algorithm and testing the robustness of MRE using the same-decision probability.
• Chapter_5: code to showcase the XBN R package developed using benchmark Bayesian networks included.
• Chapter_6: code to demonstrate the practical insights obtained from explanation of evidence and explanation of decisions.