Constructing interdependent risks network of project portfolio based on bayesian network

The risk of a project portfolio is investigated using a new methodology, which is applying Bayesian network structure learning to construct the interdependent risks network. In particular, to overcome the drawback of greedy research algorithm with random start, the mutual information between project risks is measured before running the algorithm by computing the impact of technical interactions on risks between projects. It is proved that he preprocessing can reduce the search space and exclude the error edges effectively. After comparing the performance of our method with the random greedy algorithm and the results of our method in different sample sizes, the conclusion is that our method can efficiently and accurately identify the interdependent risks network structure of a project portfolio from training data.