受到全球電信事業自由化及國內電信市場在1997年全面開放行動電話通信業務的衝激，市場競爭的型態受到大幅改變。在多家業者加入競爭及消費者對無線通信服務需求熱切下，使得電信產業呈現一片蓬勃景象。然而，隨著行動電話市場的飽和，各家業者均面臨客戶嚴重流失問題。 為了能維持市場佔有率，各業者無不卯足全力，提供各種促銷方案來吸引客戶，然而前提就是要先做好市場區隔，以往在消費者分群的研究上，大多是將有相似消費行為的顧客來作為分群的根據，但顧客的消費行為，常會因為不同時期而有所變動，所以在消費者分群上，也必須常常跟著變動，而以往的作法是每當有消費者的行為改變，就必須將全部的消費者重新分群，不論變動的人數是多是少，因而常會耗費龐大的成本，因此本研究提出一個新的演算法single-link incremental algorithm，能夠視實際變動情況來調整群內變動的資料點，而不用將全部的資料點重新分群，不但能節省龐大的成本，又能藉由正確的分群資訊了解每一群消費者的真正需求，提高顧客對促銷方案的接受度。 Since 1997, the mobile telecom industry in Taiwan has expanded rapidly to a liberalized environment; the mobile telecom market has drastic competition. Nowadays as the mobile telecom market is being saturated, those carrier companies are confronted with high customer churn rate. For reducing the customer churn rate, every telecom company try to provide many promotional plans. But it is difficult to choose suitable plan for customers who change their purchasing behavior quickly. Sometimes, only few customers changed, but we have to re-cluster the whole customer data. It is costly and inefficient. In this paper, we introduce a new cluster algorithm, called single-link incremental algorithm. It only to adjust data objects for real change situation in intra-cluster which not to re-cluster again for all data objects. Using this new method, single-link incremental algorithm, not only can reduce cost, but also gain the right market segmentation information immediately.
