隨著貨幣貶值、通貨膨脹等問題,現代人開始有了投資理財的觀念產生,但盲目的投資往往會造成資金的縮水,故專家理財因此而蓬勃發展。本研究先以統計軟體PASW (Predictive Analytics Software)分析此一資料集再以資料探勘分類技術中的決策樹應用於投資理財之電話行銷結果預測,其資料集經過分析後可預測在哪些情況下顧客較有可能做投資理財,並希望能進一步發現其中之原因。本研究採用UCI資料庫之銀行行銷資料集(Bank Marketing Data Set)總共有4119筆樣本資料,21個資料屬性,分類方法是以Weka資料探勘軟體的NB Tree、J48、貝氏網路(Bayes Net)、純樸貝氏法(Naïve Bayes)以及多層感知機(Multilayer Perceptron)共五種分類演算法。進一步地,本研究採用模糊理論來建置模糊邏輯專家決策系統,將模糊專家系統作為判斷顧客是否投資理財之用,藉由投資理財關鍵屬性來建立完整的模糊規則知識庫,讓使用者輸入對應的屬性,找出符合的模糊規則,最後預測出做投資理財的可能性為何。本研究結果可以幫助銀行經理人/專員了解顧客是否會選擇投資理財,發掘顧客對公司的潛在價值。
With the depreciation of currency and inflation, modern people have started to form notions of investment. An aimless investment often ends up losing money; expert financial management, therefore, begins to flourish. This research use Predictive Analytics Software (PASW) to analyze data. Decision tree in data miming is also used to categorize, analyze and predict the result of telemarketing. The data, after careful analysis, can help predict whether a customer is likely to make an investment. The data number collected from Bank Marketing Data of UCI is 4119, and is sub-categorized into 21 attributes. To classify those data, this study uses the five analysis algorithms: NB Tree, J48, Bayes Net, Naïve Bayes, and Multilayer Perceptron. Furthermore, fuzzy logic is used to establish an expert decision system to judge whether a customer will invest. Key attributes are used to build up a complete fuzzy rule base. When the user keys in related attributes, it can get the corresponded fuzzy rule, and come up with a prediction of their investment potential. The result of this research may help a bank manager or agent to see the possibility of customer investment and their still unrevealed worth.