Project portfolio selection and prioritization through supervised machine learning in continuous improvement projects

Purpose In a context of limited resources, manufacturing organizations increasingly need objective approaches to prioritize Continuous Improvement Projects (CIPs). Portfolio selection decisions are often based on managerial judgment, limiting the ability to maximize business value. The purpose of this study is to examine whether machine learning–based project selection frameworks can support CIP portfolio selection, identify the most accurate supervised ML algorithms, and explore the associated challenges and opportunities. Design/methodology/approach A single case study was conducted in an Italian manufacturing company with a decade-long history of implementing CIPs. Data is collected from past projects and complemented by expert evaluations of critical success factors. A fuzzy-based data augmentation technique was developed to address the limitations of a small data set, and seven supervised ML (Machine Learning) classifiers were compared to assess predictive accuracy. Findings The analysis shows that the multilayer perceptron achieved the highest predictive accuracy, demonstrating superior capability in forecasting project success across different return on investment categories. The study highlights that ML can effectively exploit organizational knowledge embedded in historical project data to improve portfolio selection. Nevertheless, the results also reveal challenges, particularly in handling underrepresented classes of projects and in balancing computational complexity with data augmentation. Originality/value This study contributes to the project portfolio management and operations management literature by introducing a novel methodology that integrates supervised ML with fuzzy-based data augmentation for CIP selection. It provides both theoretical and practical insights: theoretically, it advances data-driven approaches to rational decision-making in project management; practically, it offers managers a scalable and user-friendly decision support tool to prioritize projects and enhance strategic alignment.