Auditing Construction Cost from an In-Process Perspective Based on a Bayesian Predictive Model

Peipei Wang, Kun Wang, Yunhan Huang, Francis Fenn, Ian Stewart

Research output: Contribution to journalArticlepeer-review


The traditional audit process in construction cost control usually occurs passively at the end of a project life cycle. This calls for a predictive model that provides a framework assembling essential information and predicts construction cost overrun potential during project processes. Unlike previous mechanistic models that reflect the full formation mechanism, the model established in this paper features a fragmentary formation mechanism consisting of shortlisted critical factors. Factors were shortlisted by both theoretical and statistical criticality to construction cost overrun, dictating the factors to pass the initial literature review identification, expert opinion verification, and Pearson's chi-square tests of interdependence. The factor shortlist was compared with the initial long list identified from the literature to understand the longitudinal trend. The trend manifested in this research necessitated a shift of project management focus from technical difficulties to managerial issues, signaled by the shifting emphasis from contractor planning and control to client monitoring and management and from project difficulties to contract qualities. The shortlisted factors and their interrelationships together formed a fragmentary mechanism and gave the model structure, which was quantified with Bayesian belief network analysis. The model automatically can calculate cost overrun potentials with relevant input and use influence diagrams to find optimal decisions maximizing the expected values of construction cost overrun potential. The predictive model achieved an accuracy rate of 92.4%, which is much higher than that of the comparable model established with the full formation mechanism. This demonstrated that mechanistic models featuring a fragmentary formation mechanism well achieved satisfactory prediction accuracy and efficiency. Therefore, this predictive model provides a framework for project auditors and other relevant project management personnel to monitor project cost proactively throughout the project lifecycle.

Original languageEnglish
Article number04022003
JournalJournal of Construction Engineering and Management
Issue number4
Publication statusPublished - 25 Jan 2022


  • Bayesian belief network
  • Construction projects
  • Cost overrun
  • In-process audit
  • Prediction


Dive into the research topics of 'Auditing Construction Cost from an In-Process Perspective Based on a Bayesian Predictive Model'. Together they form a unique fingerprint.

Cite this