Prediction of Unit Price Bids of Resurfacing Highway Projects through Ensemble Machine Learning

Abstract

Resurfacing is one of the most common highway projects in Georgia and constitutes a large portion of the state’s highway investment every year. The value of the unit price bid is one of the leading indicators to comprehensively reflect the cost to the Georgia Department of Transportation (GDOT) for these projects. Compared with other cost index research, the changing trend and large volatility of unit price bids make the prediction more difficult. This research proposes a robust ensemble learning model to predict the value of unit price bids. Data on bidding prices for more than 1,4 projects in the past nine years, along with 57 related variables, are collected, and 2 of them are selected by Boruta feature analysis to train and test the model. The results are compared with those from a baseline Monte Carlo simulation and a multiple linear regression model. Comparison shows that the proposed ensemble learning model performs much better than any single machine learning model and the baseline models. The ensemble learning model has a mean absolute percentage error of approximately 7.56. The contribution of this research is a model that can be easily replicated and implemented. The model is applicable to different kinds of construction industry data, even with missing values. Prediction is stable and efficient compared with other models to the extent of authors’ knowledge.