台灣的電信業自從民國85年政府啟動電信自由化後，使得原本由中華電信獨佔的電信市場變成了競爭激烈的寡佔市場，在電信市場中，電信費用平均的支出上企業客戶明顯的比個人用戶高出許多。因此，當電信業者進入市場後想要增加營收最快的方法就是從企業客戶著手。然而國內各電信業者能夠掌握的客戶基本資料相當有限，除了證號、地址、服務合約等資料外，其他諸如客戶背景、人口統計變項等資料付之闕如。所以在實務的做法上，國內的電信業消費行為分析大多採用營收資料庫中的通聯記錄或是帳單消費金額的變化，發展企業客戶流失分析模型，尋找企業客戶流失的徵兆。由於企業客戶可能剛開始申裝電信業務時，同時也申請他家業者的服務，如果依據企業客戶帳單消費金額的增減變化設計客戶流失分析模型，該客戶則有很大有可能成為潛在的流失客戶。本研究以電信業者之中小企業客戶為基礎對象，收集實際資料後運用資料探勘的技巧加以分析。本研究以既有的客戶群加以分類，嘗試以規則推導方法，採用客戶行業別及電信費用等資料屬性，進行潛在流失客戶之分析診斷，協助企業客戶服務團隊有效的掌握客戶動向，進而增裕公司營收。 After our government released the licenses of telecommunication industry in 1996, the oligopoly market dominated by Chunghwa Telecom has changed into the competitive monopoly market. Generally average amount of transmission cost of enterprises is much higher than individual customers. Therefore, the fastest way to increase the profit is to target enterprise customers. However, expect the open information such as identification, address and contracts, the other data like customer background and geographic variation is very limited. Practically, most telecommunication providers develop a lost analysis model to search for the lost signal. The model is based on transactions variation of the bills from the consumption behavior data. To discover the potential lost of enterprise customers, the design of analysis model can be based on the variation of telecommunication cost. This is because a customer can apply services from two different suppliers in the beginning. In our research, we focused on small-to-medium enterprise customers, collected relevant data and then analyzed data by data mining technique. Specifically, we classified enterprise customers using rule induction based on attributes such as biz type and transmission cost. The model is suitable in diagnosing the potentially lost customer, In this way the provider service team can understand the customer movements effectively and then increase the profits.
