Abstract:
To tackle the limitation of feature weights based similarity model in searching similar cases, this paper proposes an improved fuzzy case based reasoning method by utilizing the features of fuzzy rules on capturing the domain knowledge. At first, by deciding the value space of feature variable and dividing the fuzzy regions, this proposed method directly obtains the rules from data. And then, by calculating the firing strength and combining the rules, the classifier can be obtained. Finally, the classification on unsolved cases is established by calculating the classification strength. It is shown from the experimental results on three standard data sets of UCI that the proposed method not only has short learning time, but also can achieve better classification performance using less cases.