Prediction of simulated cost contingency for steel reinforcement in building projects: ANN versus regression-based models

If cost contingency is too high, it might cause the construction project to be uneconomic, aborted, and lock-up funds not available for other organizational activities; if too low, it may result in unsatisfactory performance outcomes. To predict the simulated cost contingency (SCC), a Monte Carlo Simulation (MCS) is applied. However, the use of simulation in most cases is difficult, cumbersome and rarely used in the construction industry. The objective of the current research is to develop ANN and regression based (RB) models to predict the SCC for building projects due to the variability of steel reinforcement (SR) prices. For all the adopted scenarios in dealing with the independent variables, the results revealed that on the average, ANN models are more powerful than RB models in theoretical modeling of SCC when comparing its predicted value with its actual simulated value. On the contrary, RB models outperform ANN models from practical point of view when comparing predicted SCC with its real value. Thus, contractors are advised to adopt RB models in predicting SCC for the practicality in estimating cost contingency. However, the developed models are expected to be highly beneficial for contractors in making SCC calculations easier.

Automated summary

The objective of this research is to develop ANN and RB models to predict the SCC for building projects. The results show that ANN models are more powerful in theoretical modeling, while RB models outperform in practical applications.