網路銀行會員交易行為分析 – 應用層級貝氏模型建構

Internet banking member's transaction behavior analysis The Application of Hierarchical Bayesian Model

陳前堯(Chien-Yao Chen)

指導教授 ： 王鴻龍

國立臺北大學/商學院/統計學系/碩士(2012年)

摘要

本研究利用網路銀行會員交易資料庫，運用客戶在網路銀行上的交易明細，計算出相對與絕對的活躍度指標，藉此區分出每位顧客在網路銀行上活躍度的變化，預期更精確的區分客戶行為，規劃適當之行銷專案，來提昇行銷資源之使用效能和效率。

先運用交易資料計算出購買期間(interpurchase time)，用購買期間來衡量客戶的活躍度，進一步計算出平均購買期間及加權平均購買期間，比較平均購買期間及加權平均購買期間之間的差異可計算出顧客活躍度指標(Customer Activity Index, CAI)，而利用CAI相對活躍度指標可以判斷客戶活躍度趨勢。

在估計顧客活躍度時，另外採用層級貝氏估計的方法，層級貝氏估計除可以針對個體差異進行校正外，同時可以解決個體資料稀少性的問題，加上運用馬可夫鏈蒙地卡羅(Markov chain Monte Carlo, MCMC)方法模擬出每個客戶購買期間的分配，除可計算出每人之平均購買期間外，尚可得知每個人購買期間的變異，應用此估計結果再和前述之平均購買期間、加權平均購買期間及CAI進行分析比較，可產生客群區隔，再進行客群差異分析即可得到行銷意涵，提供行銷管理所需之重要參考資訊。

關鍵字

資料庫行銷 ； 層級貝氏模型 ； 購買期間

並列摘要

In this study, using Internet banking member’s transaction database, the use of customers on the Internet banking transaction details, calculated relative and absolute activity indicators to distinguish each customer changes in activity on the Internet banking. This study expected to more precise distinction between customer behavior, planning appropriate marketing projects, to improve the effectiveness and efficiency of the use of marketing resources.

First use of transaction data to calculate the interpurchase time, using interpurchase time to measure customer activity, further calculate average interpurchase time and weighted average interpurchase time. Comparison the difference between average interpurchase time and weighted average interpurchase time can be calculated the Customer Activity Index (CAI), and CAI relatively active indicators can judge the trend of customer activity.

Estimated customer activity, another method is hierarchical Bayesian estimation. Hierarchical Bayesian estimation can solve individual data scarcity problem and correct individual differences at the same time. Coupled with the use of Markov chain Monte Carlo (MCMC) methods to simulate the distribution of each customer’s interpurchase time, in addition to calculate each customer’s average interpurchase time, can still be learned every customer’s interpurchase time variability. The application of the results of this estimate and the aforementioned average interpurchase time, weighted average interpurchase time, and CAI, can produce customer segmentation. Then get the marketing implications, provide important reference information for marketing management.

並列關鍵字

Database Marketing ； Hierarchical Bayesian Model ； Interpurchase Time

參考文獻

陳信良(2005)，以層級貝氏統計方法建構一般化迦瑪分配購買期間預測模型，國立臺灣大學國際企業學系研究所碩士論文。

陳靜怡(2005)，購買量與購買時程雙變量之預測－層級貝氏潛藏行為模型之建構，國立臺灣大學國際企業學系研究所博士論文。

Allenby, G. M., Arora, N., and Ginter, J. L. (1998), "On the Heterogeneity of Demand," Journal of Marketing Research, 35, 384-389.

Allenby, G. M., Leone,R. P., and Jen Lichung (1999), “A Dynamic Model of Purchase Timing with Application to Direct Marketing,” Journal of the American Statistical Association, 94 (446), Applications and Case Studies, 365-374.

Allenby, G. M., and Rossi, P. E. (1999), "Marketing Models of Consumer Heterogeneity," Journal of Econometrics, 89, 57-78.