由於智慧型手機與手持裝置的快速崛起與發展，造就了週邊零組件的輕薄化，而WLCSP為有效降低封裝後體積與重量之高階封裝技術，該技術利用半導體堆疊方式，使得封裝後面積小於裸晶的1.5倍、長度低於1.2倍，但也因此獨特的堆疊型態與尺寸優勢，使得在不良品檢驗上有著難以克服的先天缺陷（層層包覆與弱視呈像），造成製造商在檢驗成本上居高不下。 本論文使用資料探勘中的主成分分析與因子分析，以及知識挖掘（KDD）的七大步驟，將漏檢資料做逐一的整理、分析、比較，並從中將同性質之因子找出，達到提升生產力、降低漏檢率與檢驗成本之目的。 故依據因子分析的結果，得到本研究之結論與建議如下： 1.漏檢因素之同性質因子為：2次複判與弱視呈像，所以利用光學檢測機的導 入，以降低此2項因子對漏檢率之影響。 2.建立時間序列之即時監控機制，以達到風險控管與漏檢率控制之目的。;Due to the rapid rise and development of smart phones and handheld devices, it pushed peripheral components to become much lighter and thinner. This condition produced a new technology “WLCSP”. WLCSP is a advance package process which used a lot of semiconductor methods to reduce volume and weight. It let the total package area less than 1.5 times and length less than 1.2 times. Although this package type can satisfy end customer’s desires, but it also makes a new issue to manufacturing company “inspection leakage”. In order to keep WLCSP’s advantages and avoid its harms, this paper used statistical methods to solve these problems. It includes 2 parts of this study “Principal Components Analysis and Factors Analysis”. Also through the 7 steps of Knowledge Discovery in Databases, we clarified and discover the homogeneity factors. Finally we used its result to create action plans for the purpose of productivity improvement, leakage rate reduction and inspection cost reduction. So we generated the conclusions and suggestions were as below: 1.Import automatic optical inspection equipment and provide the solution of problem solving for focus issue and weakness graphic picture. 2.Import time sequence for mechanism establishment of real time monitor, to meet 2 purposes “risk control and leakage rete control”.
