Combination of a Nonlinear Regression Model and Earned Schedule to Forecast a Project Final Cost

Accurate forecasting of a project’s Cost Estimate at Completion (CEAC) based on current performance and progress is one the main issues in project monitoring and control. For decades, Earned Value Management (EVM) has been proved itself as a valuable tool to fulfill this task and cost estimates calculated by its Cost Performance Index (CPI) are widely applicable for projects of any type and size. However, recent studies show that the CPI-based method may be valid only for large projects with long durations. As an alternative to the index-based method, techniques with regression analysis gained a great insight in this direction. The purpose of this work is to propose a new regression-based nonlinear CEAC methodology which integrates Earned Schedule (ES) concept to assume a project progress in calculating CEAC as early as when a project is 20 percent complete. The paper sets three objectives to achieve the research purpose: development of the new equation based on a nonlinear regression modelling and ES method; validation of the new technique through case study application; and, providing a comparison with CPI-based estimates to determine the best performing equation. Testing the prediction accuracy of the proposed and index-based formulae is performed by comparing values of Percentage Error (PE) and Mean Absolute Percentage Error (MAPE). Based on six case studies from construction industry, the comparison reveals that the new methodology generates better estimates (MAPE=2,88 percent) than those calculated by traditional index-based equation (MAPE=9,98 percent)