結合可能性資料包絡分析與聚類分析進行供應商評選

Suppliers Selection Using Possibilistic Data Envelopment Analysis and Clustering Analysis

為確保有高品質、低成本且可信賴的物料供應來源，供應商評選因而成為採購管理策略面的重要議題。本文在探討以可能性資料包絡分析進行供應商評選的合理程序。詳言之，可先利用可能性資料包絡分析找出供應商的可能性交叉效率指標與可能性離群指標，藉以評估供應商的技術績效與發展方向合適性。接著，建立供應商在這兩個指標上的接近關係，藉以進行聚類分析，找出供應商最適當的分群。最後，對各群供應商進行分析，並在考慮需求數量的情況下，找出一個交叉效率高而離群值低的供應商組合做為最終決策。採用此程序進行供應商評選，除可處理具模糊性的量測資料外，還可進一步利用差額變數分析結果提供獲選供應商績效改進的參考方向。

To ensure a high-quality, low-cost and reliable source of material supply, the trend of contemporary supply management is to use fewer but reliable suppliers, and maintain long-term partnership with the selected suppliers. Therefore, the suppliers selection becomes an important strategic issue in purchasing management. In this paper, how to select suppliers using the possibilistic data envelopment analysis (DEA) is discussed. Specifically, at first, the possibilistic cross-efficiency index and the possibilistic Maverick index are used in evaluating the technical performance and the suitability of developing direction of a supplier, respectively. Then, the suppliers are clustered into groups using a proximity relation, created from the distances of the suppliers measured on these two dimensions. Finally, under the consideration of demand, a set of suppliers with high cross-efficiency and low Maverick index is suggested as the final decision. In addition to the ability on dealing with fuzzy data, this approach can also provide suggestions for further improvement by using the slack analysis in DEA.