Artificial Neural Network Models: Conceptualized and Improved Method of Contingency Assessment for Nigerian Building Projects

Abstract

Precision in estimating the contingencies in project parameters has been difficult in Nigeria despite the numerous available computational techniques. This study aims at exploring the predictive ability of artificial neural network with a view of designing prediction models and ascertaining efficiencies. Secondary data from a recently completed research based on projects in the North Eastern Nigeria were analyzed. Their surveyed construction cost ranged between N103.80 million and N303 billion, while delivery periods were between 20 and 192 weeks. Others are high cost and duration drivers, which formed the artificial neural network (ANN) models input variables, 210 cost and 199-time data sets formed the output variables. Mean contingencies of 6.46% and 7.06% for cost and duration were established from the developed models. The study concludes that the developed ANN forecasting models can aid the construction contractor in estimating project contingency of higher precisions. It, therefore, recommends the conversion of the developed models into Dashboards for contractors’ use.