Assessing the predictability of construction time overruns using multiple linear regression and Markov chain Monte Carlo

This paper aims to focus on developing a baseline model for time overrun.

Information on 321 completed construction projects used to assess the predictive performance of two statistical techniques, namely, multiple regression and the Bayesian approach.

The eventual results from the Bayesian Markov chain Monte Carlo model were observed to improve the predictive ability of the model compared with multiple linear regression. Besides the unique nuances peculiar with projects executed, the scope factors initial duration, gross floor area and number of storeys have been observed to be stable predictors of time overrun.

This current model contributes to improving the reliability of predicting time overruns.