Estimation of Risk Contingency Budget in Projects using Machine Learning

To manage risks against unexpected cost overruns, project teams use Contingency Budget (CB). Its accurate estimation has been a subject of multiple studies proposing either deterministic or probabilistic models. In this study, we propose a deterministic Machine Learning-based approach to estimate CB. Based on the k-means clustering, our model integrates the Expected Monetary Value (EMV) method and binomial distribution concepts. We test our methodology using 20 risk registers containing 25 risks with associated probabilities and impacts. Using Monte Carlo simulation, we compare our model's estimates with the ones by the traditional EMV. The model provided more accurate CB estimates and is more straightforward in use than the Monte Carlo simulation.