A robust cost estimation framework in construction projects by considering cost development

Abstract

In construction projects avoiding cost overruns depends on a decent estimation of cost for the projects in their initial phase. However due to limited project information in the early phases of projects the degree of uncertainty is high and therefore cost estimation is a challenging task and today many projects still face major cost overruns.The idea behind this thesis is to propose a robust cost estimation framework that takes both managerial aspects of uncertainty management and technical methodologies of conceptual cost estimation technologies into account. Especially artificial intelligence as a decision support system for assisting the decision makers to estimate the cost with a less degree of uncertainty, whereas the magnitude and number of projects with cost overruns would be reduced.The framework asserts that cost development in different phases of project life cycle and each decision gate can be learnt by artificial neural network, and a case-based reasoning system can assist the artificial neural network to choose the most similar case in its training data set.A case study confirmed the assertion of the thesis framework, and the results show that the applicability of artificial neural network in learning complex and nonlinear relationship between inputs of it network which are ?the project phase? and ?the degree of cost development? in that phase, with even very limited training data set is very promising, therefore we can expect that by providing a well documented and structured historical project data in the case based of Case-based reasoning system the accuracy of estimation would increase.