Study on case retrieving in case-based reasoning based on grey incidence theory and its application in bank regulation
Chuanmin Mi, Hanchong Qian, Sifeng Liu, Zhansheng Chang · 2008
A critical issue in case-based reasoning (CBR) is to retrieve a usefully similar case to the problem. There are three approaches to case retrieving: nearest-neighbor, inductive, and knowledge-guide. This article utilizes a hybrid approach using grey incidence theory to case-based retrieval process in an attempt to increase the overall classification accuracy. We propose a new approach based on the absolute degree of grey incidence to calculate the degree of nearest-neighbor matching, use grey incidence order to priority analysis. The case which has the highest degree of grey incidence is the nearest neighbor case to the input case. When there is no case whose degree of grey incidence is higher than others on all attributes, we propose using analytic hierarchy process (AHP) to calculate the weight of every index, calculating the integrated degree of grey incidence (weighted sum) to find the nearest-neighbor case. Thus a new frame-work of CBR based on grey incidence analysis is built. It is an effective method for case indexing and retrieving because it is very easy to apply and its effect is good.