透過您的圖書館登入
IP:3.14.254.228
  • 學位論文

多變量分析應用在CRM之研究

Application of Multivariate Statistical Analysis in CRM

指導教授 : 吳作樂

摘要


綜觀電子商務現有的顧客關係管理(Customer Relationship Management, CRM)推薦系統使用的技術有協同過濾法(collaborative filtering)、內容導向法(Content-based)和關聯規則法(Association Rule-based)。個人化推薦系統的過程需紀錄顧客對商品項目點選次數及瀏覽時間,加上其他關鍵影響因素,導致資料維度相當龐大,上述現行使用的技術都很不易有效處理維度龐大的資料數據。 主要以多變量統計的主成份分析(Principle Component Analysis, PCA)方法,在不損失統計效力之下,將資料維度縮減,得到維度較少的多變量統計量,再採用華德(Ward)分群法找出最合適的分群k值,其次再用傳統的K-means法驗証分群的正確性。 本研究先以多變量統計研究常用的兩組資料作主成份分析(PCA)/華德法(Ward method)驗証,再以某電子商務公司的線上CRM 資料証明其正確性及可行性。 關鍵字:協同過濾法、內容導向法、關聯規則法、主成份分析、 個人化推薦。

並列摘要


The most commonly used recommendation systems for CRM applications are: collaborative filtering, Content-based , and Association Rule-based methods.In practice, all the recommendation system needs to record and process a huge amount of customer data including their browsing time on each page and necessary personal profile data, etc. Consequently, the size of data dimension increase quickly and become a very difficult problem for implementing an effective analysis. Namely, we used Principle Component Analysis (PCA) to reduce data dimension without losing its statistical relevance.Once the dimension has reduced , a hierarchical clustering method ( Ward method) is applied to get the right number of the clusters. The cluster number is also double checked by the more traditional K-means cluster method . We implemented the proposed method on two set of open source data to verify its Effectiveness. Also, a set of real word CRM data is used to demonstrate the superiority over traditional methods. Keywords: Collaborative Filtering , Content-based recommendation, Association Rule-based recommendation, Principle Component Analysis, personalized recommendation

參考文獻


12. 姚吉峰,以關聯分析及模糊分割法建構分類規則應用於CRM資料分類,國立成功大學資訊管理碩士論文。
13. 廖婉菁,民91,應用協同過濾機制於商品推薦之研究,中原大學資訊管理碩士論文。
14. 賴冠宇,民95,顧客關係管理系統的使用現況與影響因素之探索性研究,國立中央大學資訊管理碩士論文。
15. 張毓倫,民92,個人化顯隱性知識推薦方法之研究,國立成功大學資訊管理碩士論文。
17. Berelson, B., “Content analysis in communication research,” Glencoe, I11:The Free Press, 1954.

延伸閱讀