Labor Productivity Losses across Construction Trades: A Machine Learning Approach

Labor productivity is a major concern in the construction industry. Existing research on construction labor productivity (CLP) within specific trades has produced inconsistent findings due to differences in the factors analyzed. This lack of consistency makes it difficult to identify the most critical drivers of productivity losses across trades. To address this gap, this study adopts a cross-trade analytical approach to systematically identify and evaluate the inefficiencies impacting labor productivity in multiple construction trades. Specifically, the study (1) identified common organizational and project-level inefficiencies that influence labor performance; (2) conducted an expert-based survey to measure the frequency and perceived impact of these inefficiencies across key trades; (3) developed a series of extreme gradient boosting (XGBoost) models—one for each trade—to explore the relationship between specific inefficiencies and labor productivity losses; and (4) utilized Shapley additive explanations (SHAP) to interpret the models and identify the most influential productivity-reducing factors. The performance of the XGBoost models was benchmarked against four widely used machine learning algorithms: artificial neural networks (ANN), decision trees (DT), random forest (RF), and gradient-boosted decision trees (GBDT), confirming the robustness of the chosen approach. The results reveal that productivity is trade-sensitive, with different trades facing distinct challenges. For instance, the most critical factors for concreting were “decrease in the proportion of direct work” and “lack of a labor employment system,” whereas ironworking was most affected by “drawing errors/lack of drawings” and “high turnover rate.” Based on these findings, targeted strategies were proposed under five core themes: (1) labor employment terms, (2) safety culture, (3) skill development, (4) communication, and (5) quality and location of resources. Ultimately, this study provides both trade-specific insights and a holistic perspective on labor productivity, offering practical guidance for informed decision-making in light of ongoing skilled labor shortages in the industry.