Combining rule-based expert systems and artificial neural networks for mark-up estimation

Rule-based expert systems and artificial neural networks are two major systems for developing intelligent decision support systems. The integration of the two systems can generate a new system which shares the strengths of both rule-based and artificial neural network systems. This research presents a computer based mark-up decision support system called InMES (integrated mark-up estimation system) that integrates a rule-based expert system and an artificial neural network (ANN) based expert system. The computer system represents an innovative approach for estimating a contractor' s mark-up percentage for a construction project. A rule extraction method is developed to generate rules from a trained ANN. By using the explanation facility embedded in the rule-based expert system, InMES provides users with a clear explanation to justify the rationality of the estimated mark-up output. Cost data derived from a contractor' s successful bids were used to train an ANN and, in conjunction with a rule-based expert system, select the expected mark-up for a project. The combination of both ANN-and rule-based expert systems for estimating mark-up allows significant benefits to be made from each individual system, such as understanding why and how the estimated mark-up was derived and also the effects of imposing rules and constraints on a company' s mark-up estimation. The mark-up decision support system presented can assist contractors in preparing a rational mark-up percentage for a project. Moreover, InMES as proposed will assist contractors in their tender decision making, that is, whether or not to submit a bid for a project considering the estimated mark-up.