透過您的圖書館登入
IP:3.145.52.86
  • 學位論文

以資料探勘發掘行動加值服務市場之消費者特徵

Discover Consumers Patterns Using Data Mining in Mobile Telecommunication Market

指導教授 : 朱文禎
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


集群分析於市場區隔是種很常見的方法,有統計集群方法以及類神經網路兩種方法,本研究比較自組織映射圖網路(Self-Organization-Map,SOM)結合K平均法(K-means method)、SOM、華德法、兩階段集群法,選出兩種集群方法績效較佳的方法再結合決策樹和關聯法則應用於第三代行動通訊(The 3rd Generation Mobile Telecommunication﹐3G)加值服務市場,並以加值服務在消費者心中之重要度、使用動機和生活型態劃分市場區隔找出消費者特徵。 研究結果發現,SOM結合K-means與兩階段集群績效較佳。並且在SOM結合K-means以及決策樹和關聯法則的「應用娛樂型」與兩階段集群結合決策樹和關聯法則的「社會務實型」都有發現到高度偏好生活應用及娛樂之加值服務,以及前方法的「簡訊務實型」與後方法的「親友務實型」都高度喜歡簡訊與監控之加值服務,使用動機高度受到生活需求的影響。 本研究發現透過決策樹分析後能得知消費者特徵範圍的偏好,而且關聯法則能夠探勘到決策樹分析難以發現到的人口統計以及手機使用狀況。由此可知整合不同集群方法來進行市場區隔,可以加深對各個市場區隔的瞭解。

並列摘要


Clustering analysis is a common method for marketing segmentation, including multivariate analysis and artifical neural network. This study aims to compare four clustering methods︰(1)intergration of SOM(Self-Organization-Map) with K-means, (2)the SOM, (3)the Ward’s, (4) the two-stage . Select the best two of them to combine with decision trees and association rule for applying in the 3G mobile (The 3rd Generation Mobile Telecommunication) value-added services which divides each segmentation with service attribute, consumers motivation, and lifestyle to discover consumer characteristics. The results show that method(1) and method (4) are better performance than other methods. Moreover, the “Application-oriented” of method (1) with decision trees and association rule and “"Realistic oriented” of method(4) with decision trees and association rule both are highly prefer life applications and entertainment service. “Message-oriented” of method(1) and “Emotion-oriented” of method(4) both highly use message and monitoring services as well as consumers motivation is highly affected by life needs. The paper finds out the scopes of consumers’ characteristic through decision trees. The association rule is able to dig out the demographic data and mobile usage situation , whereas the decision tree can not find out. Therefore, integration of different clustering methods for marketing segmentation can help comprehend each segmentation more detail.

參考文獻


3.朱文禎、駱明熙(2006),行動電話號碼可攜服務開放後的用戶市場區隔探討-以台南地區為例,電子商務學報,4(4),pp:379-404。
9.張心馨、蔡憲富(2004),“以Data Mining技術結合SOM和K-means的消費者集群方法與顧客關係管理和績效獲利性評估之實證研究”,《資訊管理學報》,11(4),pp:161-203。
23.郭英峰、游景文(2007) ,“消費者採用行動加值服務行為意向之研究-以年輕族群為例” ,《資訊管理學報》,14(3),pp:125-153。
28.黃淑姿、李冠穎、許英傑(2010) ,行動加值服務價值創造對行為意圖影響之研究,電子商務學報,12(1),pp︰41-72。
16. Lien, C. , Ramirez .A. and Haines. G. (2006), “Capturing and Evaluating Segments: Using Self-Organizing Maps and K-means in Market Segmentation”Asian Journal of Management and Hummanity Sciences, 1(1), pp:1-15.

延伸閱讀