應用資料探勘技術探討顧客保留模型

The Application of Data Mining Techniques in Customer Retention Model

劉彥辰(Liu Yen-Chen)

指導教授 ： 施人英

國立臺灣師範大學/管理學院/全球經營與策略研究所/碩士(2015年)

https://doi.org/10.6345/NTNU202205449

摘要

直效行銷近年來在物流發展與法規的保障等因素下已經有明顯的成長，各大通路業者拜資料庫系統健全之賜，透過來往交易中得到豐富的顧客資料與交易紀錄，如性別、年齡、購買金額、購買時間、購買商品類別等，因此如何妥善利用這些資料來提升對顧客的了解，並使用適當的資料探勘技術，提供給顧客個人化的行銷與服務，讓通路業者能在競爭中獲得先機，將成為未來廠商重要的課題。
本研究利用資料探勘中集群、分類、關聯分析技術於顧客關係管理，達到顧客保留之目的。將透過一個年度消費者交易資料及商品資料，轉換出176筆攸關變數，先將消費者依照購買行為做出分群，再運用決策樹、類神經網路與羅吉斯迴歸作為分類分析的工具，尋找出下一個年度第一期型錄寄發名單中的消費者是否購買商品，並評估分類之準確度，再比較未經分類直接採用各分類分析的準確度。透過關聯分析中將顧客分為男女兩組，探討顧客在商品購買中的關聯性，結合分析結果，找出目標消費客群，並期望能將此資料探勘之成果研擬出針對不同顧客的行銷策略，達到顧客保留之目的。
透過本研究之分析，通路廠商能夠更有效對顧客進行區隔，根據顧客之消費特性及基本資料所產生的變數列出分類規則，能夠在型錄寄發對象的選擇有更客觀的參考依據，而研究成果期待能為通路廠商將有限的行銷資源做出最有效率的發揮。

關鍵字

資料探勘 ； 顧客關係管理 ； 顧客保留 ； 決策樹 ； 類神經網路 ； 羅吉斯迴歸

並列摘要

In recent years, direct-marketing has grown significantly based on the evolvement of logistics and the rule of legal protection. Most of stores developed database system to collect customer profiles and transaction records. The most important issue in these stores is how to use the information to enhance realization of customer transactions properly. We can use data-mining technology to analyze these data and provide customization service to customers.
This research used data-mining (clustering, classification and association) in customer relationship management to reach the target customer retention. The customer profiles and one-year transaction records were used to generate 176 variables. The first step is to conduct cluster analysis based on purchasing behavior of consumers then classification analysis is used in each group of customers to find out who will buy through catalogue in the next year. The second step is to compare the precision of the model without using cluster analysis. Finally, we use association to explore customers’ purchase association among items. We integrate the results of analysis to find target customers and develop marketing strategies for customer retention.
Through this research, companies can sell products effectively. According to the classification rules, companies can choose the right customers to send catalogues objectively. The expectation of this research is to let the sellers maximize their efficiency under the condition of limited marketing resources through data mining.

並列關鍵字

data mining ； customer relationship management ； customer retention ； decision tree ； artificial neural network ； logistic regression

參考文獻

Agrawal, R., Imieliński, T., & Swami, A. (1993). A Novel Genetic Algorithm Based on Image Databases for Mining Association Rules. ACM, 207-216.

Agrawal, R.,& Srikant, R. (1994). Fast Algorithms for Mining Association Rules. UMI Dissertations Publishing.

Anderberg, M. R. (1973). Cluster Analysis for Applications. Academic Press.

Berkson, J. (1944). Application of the Logistic Function to Bio-assay. Journal of the American Statistical Association, 9, 357-365.

Berry, M.J.A., & Linoff, G. (2004). Data mining techniques: for Marketing, Sales, and Customer Relationship Management. Wiley Pub.