Scatter degree-based case retrieval method for CBR in project cost assessment

To improve assessment efficiency of construction project cost and enhance the development of electronic commerce for real estate, case-based reasoning (CBR) has been adopted by us. However, when a new cost assessment problem is input into the CBR system with the nearest neighbor method, there are usually little difference on similar degree for the retrieved most similar cases, thus, it is usually difficult to decide which case to be adapted so as to solve new assessment problem. Thus, the scatter degree method is introduced into the case retrieval. Some disadvantages of the conventional scatter degree method is analyzed, the rank correlation method and genetic algorithm (GA) are introduced to adapt the scatter degree method. Then the adapted scatter degree method is used to determine the weighting coefficient of each case feature. The scatter degree-based case retrieval method is adopted the practical cost assessment of construction project, and the result validates the validity of above model. It demonstrates that the scatter degree-based method is more efficient than the conventional nearest neighbor method, and the solution precision can be improved by the degree-based method.