Improving preliminary cost estimation in Indonesia using support vector regression

Abstract

A typical preliminary cost estimation for a building construction project in Indonesia may take weeks and have an error rate varying from around −13 to +27%. The research objectives of this study were thus to determine the factors that influence cost estimation in Indonesia and to develop a support vector regression model to improve the accuracy and reduce the work hours for preliminary cost estimation. The literature review identified 14 factors that have the most influence on cost estimation in Indonesia. Data collection was carried out to gather information randomly on 104 building cases in Indonesia deemed to contain valid information for the proposed model. The model with a radial basis function kernel was established after data trimming, analysis and normalisation. The proposed model was then evaluated and implemented using fivefold cross-validation and yielded an average accuracy of 96% for preliminary cost estimation of building construction projects. The model’s average accuracy was 87%, it was efficient and resulted in significant time savings.