시계열 군집화를 반영한 선형혼합모형 기반 EVMS 최종사업비 예측 연구

Abstract

This study proposes a novel approach to predict the estimate at completion (EAC) within earned value management systems (EVMSs) using a l inear mixed model (LMM). Unlike conventional methods that rely significantly on subjective judgments, this model incorporates statistical techniques to enhance the objectivity and accuracy of cost estimation. The model includes both fixed effects, such as planned value (PV), schedule performance index, and time, as well as random effects based on project clusters derived from dynamic time warping (DTW) analysis. The model was validated using data from 30 completed projects and further validated using new project data. The results indicate that the proposed model consistently outperformed conventional methods, with prediction errors within ±5% of the actual final project costs achieved. This approach effectively captures the temporal and project-specific characteristics, thereby providing a robust tool for early-stage cost predictions. The findings underscore the potential of the LMM-based EVMS in supporting decision making and optimizing resource management across diverse project types.