為因應經濟體系之全球化與國際化，為能快速回應顧客需求與有效協調資源供給，導入企業資源規劃 (Enterprise Resource Planning, ERP) 系統已成為企業提升競爭力的關鍵要素。然而ERP 系統導入過程複雜且成本昂貴，倘導入失敗，恐致企業產生營運危機，故在導入之前，需要格外慎重評估。本研究採用「國立中央大學ERP 卓越計畫研究中心」之問卷資料進行分析。首先，藉由文獻探討歸納出企業導入ERP 軟體評選時應考量之評選準則，包含：成本與營運效率、系統支援能力、系統相容程度與彈性、流程適切性及系統功能完整性;其次，應用自我組織特徵映射網路(Self-Organizing Feature Map Network, SOM)與K-means cluster法來分群，最後，藉由貝氏網路，分析兩者分群結果。研究分析獲得系統相容程度與彈性，為最重要評選準則，次為成本與營運效率、系統功能完整性、系統支援能力與流程適切性。另分群準確率而言，K-Means cluster 法以99.65%高於SOM 法97.88%；不過就資料收集與分析成本效益，SOM 法可省84%優於K-Means cluster 法57%。本研究的結果，可以提供企業評選ERP 系統之參考。

During the highly changed environment, it is the critical factors to promote theenterprises competitiveness by importing ERP system. However, it takes a great deal ofresources to develop ERP system. It has to carefully evaluate before installing ERP system.Otherwise, it would lead to operational crisis for enterprises.This study references the questionnaire survey data from the ERP Excellence ProjectResearch Center of National Central University. It is subjects of the investigation for thoseenterprises imported the ERP system. It is going to clustering by using SOM and K-Means infirst phase. There are cost and operational efficiency, system support ability, systemcompatibility and flexibility, process appropriate and system functional complete as inputparameters. In second phase, Bayesian Network analyzed the accuracy of clustering for abovetwo methods. System compatibility and flexibility are the most important selection criteriafactors from the study result. For the accuracy of clustering, the K-Means 99.65% is higherthan the SOM 97.88%. But for information collection and cost effectiveness analysis, theSOM 84% is higher than the K-Means 57%. The result of this study can be the reference forthe enterprises installing ERP system.