由於消費性電子產品獲利降低，許多公司競相投入獲利較高的工業電腦產 業。除提供滿足客戶需求的創新設計與應用外，業者同時要維持良好品質與效 率，以擊敗競爭對手及確保利潤。本研究首先彙整某工業電腦公司製造執行系 統 (manufacturing execution systems, MES) 與退貨商品授權 (return merchandise authorization, RMA) 中的品質異常資料；接著，應用Apriori 演算法找出 RMA 與MES 間的品質異常關聯並建立RMA-MES 預測模型，以預測製程異 常所可能造成的退貨問題；最後，分析與討論RMA-MES 關聯資訊，以做為 RMA 及MES 發生異常時，管理人員採取適當對策及防禦措施之參考。

Due to the low profit return of consumer electronic products, many companies transferred their investment to the industrial personal computers (IPC) industry to pursue higher profit. In addition to providing innovative design and application to meet customer requirements, manufacturers also need to keep good quality and high efficiency to defeat competitors and ensure profits. This research first collects the abnormal quality data from the Manufacturing Execution System (MES) and Return Merchandise Authorization (RMA) of a real IPC company. Then, the Apriori algorithm is applied to analyze the association of quality abnormalities between RMA and MES, and further, to build the RMA-MES forecasting model to predict the possible return merchandise problems from the manufacturing process abnormalities. Finally, RMA-MES association information is discussed and explained to provide managers a reference for taking proper countermeasures and prevention when RMA and MES abnormalities occur.