Construction project control is generally reactive, often resulting in schedule and cost deviations, often resulting in delays, and cost overruns. This study aims to formulate a business intelligence framework for weekly project control, develop a predictive XGBoost model for early warning, and develop lean construction-based corrective action recommendations. A quantitative approach was used on PT Royal Project Management's Type 45 house project, utilizing weekly data in the form of schedules, RABs, WBS-based physical progress, and actual costs. Data processing was performed through earned value management, S-curve analysis, time-based split analysis, and walk forward validation. The results show that integrating PV, EV, and AC into a BI dashboard produces consistent performance indicators. The XGBoost model also performed better than the baseline, with a predicted MAPE of 3.81% for EV and 1.16% for AC. The predicted outputs were then converted into normal, alert, and critical risk statuses and linked to lean-based SOPs to accelerate managerial response.
Copyrights © 2026