Refinement of weights using attribute support for multiple attribute decision making

A number of approaches have been proposed to determine the weights for multiple attribute decision making. However, the resultant weights are usually assumed to be fixed, making it lack of tolerance to accommodate variation if the patterns of the subsequent data are subject to change. This article proposes a method to facilitate the adjustment of attribute weights, which accommodates a number of relevant characteristics. A model is first constructed that is able to express the requirements of a particular application. The concept of attribute support and consensus are then proposed for subsequent weight modification. A full algorithm is finally presented for the attribute weight adjustment. The effectiveness of the proposed method is validated by way of a case study in the tax credit domain with a sensitivity analysis of the method further evaluated.