Abstract:Benchmarking , which is one of the methods for supply chain assessment , has important application value ,while how to select appropriate benchmark is the bottleneck of this method. Aiming at this problem , this paper dis2 cusses how to collect and analyze the characteristics of supply chains with density-based clustering mining technolo2 gy , so that it can provide decision support to compare and improve supply chain performance with benchmarking. In the approach , firstly , the index values of supply chain performance are standardized. Then , they are classified by density-based clustering technology (improved K-average clustering method) . After the analysis of each cluster , the problemof benchmark selection is solved. Finally , a numerical example is given to illustrate the proposed approach