Enhancing Project Portfolio Risk Response Decision: A Distributed Approach Considering Interproject Failure Correlation
提出一种双层分布式方法，利用贝叶斯网络分析项目间失败相关性，优化项目组合风险应对决策，并通过案例验证其有效性。
An effective Project Portfolio Risk Response Decision (PPRRD) is crucial for enterprise strategy. Notably, inter-project failure correlation can propagate Project Failure (PF) risks across the Project Portfolio (PP), thereby complicating PPRRD. To address the challenge, a novel two-layer distributed approach for PPRRD is proposed. First, the Bayesian Network (BN) measures correlations between risks and the spontaneous probability of PF to analyze inter-project failure correlation and Project Risk Response Decision (PRRD). Second, an input-output model evaluates the integrated probability of PF and cascading effects among projects, formulating an upper-layer risk response resource allocation model for the PP. Third, constrained by risk response resources, a lower-layer PRRD model is developed. The integration of upper-layer and lower-layer models forms the comprehensive two-layer PPRRD model. Finally, research rigor is demonstrated through a real case study, benchmarking against a centralized baseline and an independence-based setting, and conducting sensitivity analysis. Results show that: (1) the distributed approach maintains solution optimality relative to centralized methods; (2) better results can be obtained considering inter-project failure correlation; (3) Project Portfolio Decision-maker (PPDM) should avoid overly strict constraint optimization; (4) more attention should be paid to projects characterized by both high spontaneous probability of PF and dependence degree. This study contributes a new decision making framework that combines inter-project failure correlation, offering practical insights for PP risk management under uncertainty.