PROJECT PERFORMANCE PREDICTION BASED ON MACHINE LEARNING: A SYSTEMATIC LITERATURE REVIEW

Abstract

Context: For successful project management, it is crucial to accurately predict project performance, as this enables effective resource allocation and proactive decision-making. Traditional approaches often fail to achieve this effectively, leading to increased interest in using machine learning techniques due to their reliable predictive capabilities. Objective: This research study conducts a systematic literature review on predicting project performance to examine the current studies on machine learning applications. Methodology: This systematic review employs the PRISMA technique to analyze 34 relevant studies from the Scopus, IEEE Xplore, and PubMed databases, published between 2015 and 2024. This study presents three research questions that examine trends in project performance based on machine learning studies, identify the machine learning models most frequently used in the context of project performance, and explore their applications in various project contexts. Findings: The findings indicate that research trends increased from 2020 onwards, with support vector machines and artificial neural networks being the most commonly used models among machine learning models in the context of project performance. It also found that most research conducted in the software and construction industries utilized time and cost-related features, with only one study observed in the context of a student project. Conclusion: The findings of this study confirm that the machine learning approach has strong potential to increase prediction accuracy in project management across different project contexts. This systematic review presents a comprehensive taxonomy of machine learning techniques and identifies key research gaps. It highlights the need for validating machine learning models in real-world contexts and also introduces hybrid models that can integrate human expertise. This study is a valuable source that may help practitioners and researchers leverage machine learning in the project domain.