NEURAL NETWORK APPLICATION FOR BUILDING PROJECTS COST ESTIMATION

Abstract

This work presents a neural network based cost estimating method, developed for the generation
of conceptual cost estimates for total building and electromechanical systems in building project,
by using eight parameters available at the early design phase. This model establishes a
methodology that can provide an economical and rapid means of cost estimating. Eighteen ligh
rise building projects, built between 1996 and 2009 in Middle East countries used in this study.
The performance of developed cost models was tested against costs incurred by projects not used
in training of those models. Results show the mean absolute percentage errors (MAPE) are
between 1.51% and 4.771 % for the five networks, and the maximum/minimum deviation of the
cost estimation is 10.2/0.17. These figures considered good cost estimation at the early design
stage.