Discriminant analysis for predicting ranges of cost variance in international construction projects

Research output: Contribution to journalArticlepeer-review

46 Citations (Scopus)


Unanticipated market conditions as well as project-related risks can easily lead to cost overruns in international construction projects. For a contractor to be financially successful in international projects, a careful examination of the project is a prerequisite to understanding the cost variance characteristics. Based on the reasonably accurate characterization of the cost performance, the markup or contingency amount is determined to ensure both a decent level of profit and a good chance of winning the contract. This paper presents a classification model to categorize international construction projects, particularly faced by Korean contractors, into five cost-variation classes: extreme cost overrun, moderate cost overrun, neutral, moderate cost saving, and extreme cost saving. The model is able to characterize an international project for its cost performance prediction in comparison to the contractor's initial cost estimate. A linear discriminant analysis is utilized to develop the predictive classification model with the support of the bootstrap method. Tests show that the proposed model is able to help cost estimators determine a proper level of cost contingency before bidding on an international project.

Original languageEnglish
Pages (from-to)398-410
Number of pages13
JournalJournal of Construction Engineering and Management
Issue number6
Publication statusPublished - 2008

Bibliographical note

Funding Information:
The authors thank Shell U.K. Ltd. and its partner, ExxonMobil, for their support and permission to publish this paper.

All Science Journal Classification (ASJC) codes

  • Civil and Structural Engineering
  • Building and Construction
  • Industrial relations
  • Strategy and Management


Dive into the research topics of 'Discriminant analysis for predicting ranges of cost variance in international construction projects'. Together they form a unique fingerprint.

Cite this