ISP Customer Relationship Management using Visualization Data Mining

[[abstract]]隨著電信自由化， 吸引了相當多網際網路服務業者(Internet Service Provider；ISP)紛紛投入市場，因此大幅影響ISP 產業的競爭局勢。在這樣競爭 激烈的環境下，ISP 網路業者須投入相當多成本以提升網路頻寬進而滿足消費者 對頻寬的需求，而消費者偏好低廉的網路租賃費用，造成ISP 網路業者經常以低 價，甚至以不計成本的策略爭取顧客，導致顧客轉換接取服務業者的頻率升高。 因此，如何能夠有效地降低客戶流失率進而提高其忠誠度是ISP 網路業者十分重 視的課題。降低客戶流失或提供顧客忠誠度之可行手段之一，是提高服務品質， 讓客戶滿意於ISP 網路業者的服務而樂於長期使用，而有效運用顧客關係管理 (Customer Relationship Management；CRM)所產生的資訊或知識，可強化企業對 顧客個人化的服務以減少顧客流失及增加顧客忠誠度，甚至可提高顧客對企業的 價值。ISP 網路業者擁有了相當廣大的顧客，不同的顧客對於網路的需求皆不相 同，也衍生出不同的網路使用行為。若能了解顧客使用網路的行為，輔以顧客關 係管理的學理，對ISP 網路業者在市場的競爭優勢上無疑是一大助力。 本研究以國內某ISP 網路業者之顧客網路流量記錄，依循產業標準的資料探 勘方法論CRISP-DM，並結合資料倉儲之線上分析處理及視覺化資料探勘技術以 探勘顧客網路使用行為。再應用自組織神經網路及RFM 顧客價值分析模型將顧 客的網路使用行為進行分群及顧客價值的運算，協助ISP 網路業者更加的了解其 顧客的網路頻寬使用需求與顧客的網路使用行為，使ISP 網路業者與顧客的關係 更加的緊密。顧客價值的運算結果由領域專家進行分析後，全部的顧客中有 30.77%採用主動行銷策略而16.23%顧客關懷策略。網路使用行為的分群上，共 分為9 群，群與群之間可明顯的發現其不同的網路使用型態，並發現到不同的顧 客網路使用行為可能都使用主動行銷的策略，因此在主動行銷的策略上可以針對 不同的網路使用行為制定不同的策略。藉此，ISP 網路業者可適時適地的提供顧 客關懷與主動銷售等服務，以增加其獲利能力。此外，本研究也利用網路流量記 錄，進行路由器及行政區域網路流量分佈的知識發現，並利用視覺化的方式呈 現，以協助決策者在更短的時間內取得更多的關鍵性的資訊。此可協助ISP 網路 業更準確的投資網路設備，以減少營運成本。如此的循環下，勢必能讓ISP 網路 業者有效率的面對變遷與競爭的環境，使其獲得永續的生存與發展。[[abstract]]Telecommunication liberalization in Taiwan attracts a lot of Internet Service Providers (ISPs) to invest into the internet service market. The influence of the ISP industry has become more and more competitive. To satisfy customers’ need in such competitive ISP market, the ISPs must raise the network frequently wide demand and lower the price. The cutting price strategy causes customers to change the ISPs frequently. Therefore, it is important for ISPs to know how to keep their customers’loyalty in the long run. The customer satisfaction is an important strategy to keep customers’ loyalty. ISPs have to improve their service qualities and make customer satisfactions. By using the Customer Relationship Management (CRM) to strengthen individual service, it will not only increase customers’ loyalty, but also enhance the customers’ value. Different customers have different network accessing behaviors and needs. If we can understand the customers’ accessing network behaviors and the applications of CRM, it will be a powerful competition advantage for ISPs. Therefore, the purpose of this study is to understand customer behavior of accessing network, and to provide decision makers to get key information quickly. First, this study utilized the Multi Router Traffic Grapher (MRTG) Log of customers in one of the Taiwan’s ISPs. To understand customer behavior of accessing network, we followed CRoss Industry Standard Process for Data Mining (CRISP-DM) methodology to combine On-Line Analysis Process (OLAP) and visual data-mining. In addition, to support ISPs understanding network bandwidth utility rates, we applied Self-Organizing Map (SOM) and RFM Model (Recency, Frequency, and Monetary) to cluster customers and to calculate customer values. ISPs can actively provide customer cares and services to increase their profit. Furthermore, to provide decision makers the visualized results of data-mining and to get key information quickly, we performed data-mining in router and administrative divisions of distribution of network traffic. Therefore, ISPs can invest more accurately in network equipment and reduce their operation costs. In the long run, ISPs can face the competitive environment change and survive strongly