Off-site steel construction (OSC) is an emerging alternative to traditional on-site methods that offers advantages such as reduced waste, improved quality, and faster delivery. However, OSC also requires robust monitoring and control mechanisms within fabrication shops to ensure the timely and cost-effective completion of projects. This study presents the development and application of a machine learning (ML) model to estimate the cost performance index (CPI), a pivotal indicator of project progress and performance, by considering influential factors that affect OSC projects. A comprehensive analysis of data from 56 OSC projects fabricated in a steel parts manufacturing facility between 2020 and 2022 was conducted. The investigation focused on four key features: project weight, project utilization type, project connection type, and material supply method, examining their impact on CPI. The data set was meticulously preprocessed, and three ML algorithms—support vector machine (SVM), gradient boosting (GB), and decision tree (DT)—were employed to model CPI. Model performance was evaluated and compared using metrics including root mean square error, accuracy, and R-squared. The findings demonstrated that GB excelled in CPI prediction, achieving an accuracy rate of 91%. This research underscores the utility of ML as a valuable tool for monitoring and controlling off-site steel construction projects. It also provides insights into the factors that influence CPI and suggests ways to optimize them for better project outcomes. Furthermore, the study contributes to the literature on OSC by exploring the relationship between project characteristics and performance indicators, which can help practitioners improve their decision-making and planning processes. The study also discusses the limitations and challenges of applying ML models to OSC data, such as data availability, quality, and consistency.

Keywords