Estimation of building construction cost using artificial neural networks

Abstract

The cost estimation with higher degree of accuracy at initial stage of the building construction projects plays a vital role in the success of every construction projects. Based on a survey of design professionals and contractors, dataset of 78 buildings construction projects was obtained from urban city of Mumbai, India and nearby region. The most influential design parameters of the structural cost of buildings were identified and assigned as an input and the total structural skeleton cost signifies the output of the neural network models. This paper discusses the development of a multilayer feed-forward neural network model trained with a back propagation algorithm for the prediction of building construction cost. The early stopping and Bayesian regularization approaches are implemented for the better generalization competency of neural networks and to avoid the over fitting. The Bayesian regularization approach performance level during the construction cost prediction is better than early stopping. The results obtained from the trained neural network model shows that, the neural network is able to predict the cost of building construction projects at the early stage of the construction. This paper contributes to construction management and provides the idea about the entire outlay budget which will be helpful to the owners and investors in decision making and to manage their investment.