Cost overrun classification is a critical aspect of project management in Iran’s petrochemical industry, demanding accurate predictive models to inform decision-making and mitigate risks effectively. This article focuses on developing a decision tree-based predictive model specifically tailored to classify cost overrun levels as high, moderate and low. Machine learning techniques extract transparent and interpretable rules for cost overrun classification using a rule-based approach within the decision tree framework. The process involves meticulously extracting rules from the decision tree model to provide actionable guidelines based on locally relevant risk factors. The results demonstrate the model’s effectiveness in systematically assessing risk factors and accurately predicting cost overrun classifications. Thirteen rules are extracted, each constituting a unique combination of conditions leading to a specific classification outcome, enhancing the model’s validity and reliability. The successful completion of this phase delivers a valuable predictive tool for project managers and stakeholders in Iran’s petrochemical industry. The developed decision tree model allows stakeholders to navigate cost overrun challenges effectively and optimise project outcomes by enabling informed decision-making and proactive risk management.
Mohseni et al. (Wed,) studied this question.