自政府實施金融自由化政策後,國內的銀行業面臨了激烈的競爭與挑戰,而銀行營收以貸款業務為主要來源之一,本研究目標將鎖定於中小企業(Small Medium Enterprise,簡稱SME)借貸戶資料集,運用資料探勘(Data Mining)中的分類(Classification)技術,將決策樹演算法與類神經網路演算法交互使用,建立二階段的分類模型,找出有資金需求的中小企業借貸戶之特徵,藉此協助管理者制定相關的決策,發掘目標顧客的需求因素。研究結果顯示在不影響F-measure整體效能的情況下,二階段分類模型比一階段的分類模型提升了捕捉率(Recall Rate),可以探勘出更多的中小企業借貸戶。而本研究結果鎖定的中小企業借貸戶,不僅和資料集母體中會來借款的公司戶的行為相似,甚至有更大的金錢流動,更有資金上的需求。
After the government's financial liberalization policy, banks have confronted the severe competition and challenge. One of the bank's revenue is from unsecured loan. Therefore, the aim of this study focuses on Small Medium Enterprise (SME) data which has the demand feature of SME loan. The classification techniques of data mining, Decision Tree and Neural Network, are used to build the two-stage model.The results show that two-stage classification prediction model enhances Recall rate without effecting F-measure which compares with one-stage classification prediction model. Moreover, the targeted SME are not only similar to demand feature of SME in dataset but also have more cash flow and requirement of cash. It can provide decision maker for an appropriate marketing plan to the customer in the bank.