Journal of Information Technology in Construction
ITcon Vol. 19, pg. 399-411, http://www.itcon.org/2014/24
Development of a two-step neural network-based model to predict construction cost contingency
published: | September 2014 | |
editor(s): | Amor R | |
authors: | Sang C. Lhee, Ph.D., Rinker School of Construction Management, University of Florida, email: lheesch@ufl.edu Ian Flood, UF Research Foundation and Holland Professor, Rinker School of Construction Management, University of Florida, email: flood@ufl.edu Raja R.A. Issa, UF Research Foundation and Holland Professor, Rinker School of Construction Management, University of Florida, email: raymond-issa@ufl.edu | |
summary: | An owners cost contingency is one of the most important cost elements within a base estimate to account for unpredictable risks and changes in the delivery of construction projects. The accurate estimation of an optimal contingency is critical for the financial success of a project and for ensuring the optimal use of an owners funds. Existing methods for estimating contingency are deficient in that the answers they generate are often far from optimal causing problems such as depletion of budgets, disputes, and reduction in work quality. This study proposes a two-step neural network-based method for estimating the optimal contingency for an owners funding of transportation construction projects that has the objective of achieving solutions that are closer to the optimum than existing tools. The two-step method is a development of a one-step neural network approach that has been found to perform better than the approach currently adopted by the Florida Department of Transportation (FDOT). The two-step ANN-based prediction model is shown to generate estimates of contingency that are closer to the optimum than the one-step ANN-based approach. As a consequence, the two-step approach has the potential to improve an owners budgetary decisions, reducing the risk of either underutilizing or over committing funds. | |
keywords: | Artificial neural networks, Construction cost contingency, Form of contingency, One-step ANN-based model, Two-step ANN-based model | |
full text: | (PDF file, 0.962 MB) | |
citation: | Lhee SC, Flood I, Issa RRA (2014). Development of a two-step neural network-based model to predict construction cost contingency, ITcon Vol. 19, pg. 399-411, https://www.itcon.org/2014/24 |