Predicting Bidding Price in Construction using Support Vector Machine

Abstract

The bidding price can directly affect the construction firm business, so the decision what price to bid for a new project in order to sign a contract is a difficult and responsible decision. Data for fifty four tenders were collected from construction firms. Only data for twenty six tenders were contracts which were signed and used for further analysis. A model for forecasting the bidding price was developed. The model uses support vector machine. DTREG software package was used for predictive modeling. The obtained prediction was very accurate with only 2.5% error (mean absolute percentage error) of the model.