A Bayesian approach for forecasting errors of budget cost estimates

Abstract

Accurate estimation of budget costs is critical for effective management of construction projects. The performance of various management functions is dependent on the accuracy of the estimates throughout the construction phase. These estimates, however, inevitably involve a considerable amount of error, which imperatively requires the evaluation of budget cost estimates and the measurement of errors associated with the estimates. Applying an analytical procedure, this study carried out a thorough statistical analysis of existing practice in the construction industry to identify limitations of the practice. As an alternative to the practice, a Bayesian approach was found to be more appropriate than the industry common practice to account for the probabilistic nature of estimates and to forecast errors associated with the budget estimates. A scenario-based example is included to demonstrate application of the analytical procedure for analysing historical cost performance data that are readily available in most construction companies.