應用層級貝式理論於跨商品類別之顧客購買期間預測模型

Abstract

近年來，由於消費者需求日漸分歧，大眾媒體式微，利用資料庫進行一對一行銷已成為必然的趨勢，其中，顧客之購買期間變數對於虛擬通路之企業維持顧客關係相當重要。過去學術界上已經有不少以顧客購買期間為主要變數之研究，然而，跨商品類別之議題卻較少被提及；跨商品類別之顧客行為模型多半著墨在顧客購買時點(Purchase Incidence)或是品牌選擇之議題。不同之商品類別各有其購買週期，且在多商品類別的環境下，不同商品類別之顧客購買期間並不互相獨立，因此，本研究擬利用層級貝式模型可估計出顧客異質性參數之優點，建構一跨商品類別之顧客購買期間行為預測模型。 本研究根據Allenby et al.(1999)所提出之顧客購買期間模型為基礎，以Generalized Gamma分配來配適顧客購買期間變數，並利用Multiplicative Model將商品類別變數引入基礎購買期間模型當中，以建立跨商品類別之顧客購買期間預測模型。最後，利用危險率函數的概念，本研究計算出每位顧客之購買機率，並進行排序作為預測之用。 本研究利用國內一型錄購物公司之顧客交易資料進行實證研究，並將跨商品類別模型與過去之層級貝式購買期間基礎模型進行預測命中力之比較。此外，本研究亦將顧客交易資料依照購買商品類別區隔開，各自進行基礎模型之參數估計並比較預測命中力。本研究根據此實證資料可得到以下結論： 1.驗證過去以Generalized Gamma分配配適顧客購買期間的適當性 2.於此資料當中，本研究所提出之跨商品類別購買期間模型之預測命中率優於過去之購買期間基礎模型。 3.跨商品類別顧客購買期間模型可以估計出不同商品類別的購買轉換乘數，進而瞭解不同商品類別轉換對購買期間的影響程度。

Because of the recent diversity of consumers’ demand and the less popularity of mass media, one-to-one database marketing has been utilized by companies to increase their competitive capability. To maintain better customer relationship, companies such as on-line stores must understand the customers’ behavior in terms of inter-purchase time. There have been many research literatures addressing the issue of inter-purchase time; however, few of them consider the impact of multi-category of products on inter-purchase time which may very under different products. Therefore, the goal of this paper is to build a one-to-one multi-category inter-purchase time model by using the hierarchical Bayesian model. The hierarchical Bayesian model proposed by Allenby et al.(1999) has been extended based on the generalized gamma distribution and multiplicative model formulations. With the use of Hazard rate function, the model is then used to derive the purchase probability of each individual customer. To validate the proposed model, field data from a local catalog company are collected. Prediction hit rates by different models are compared. The conclusions are as follows: 1. Generalized Gamma distribution is a good and flexible distribution to model customer inter-purchase time. 2. The multi-category inter-purchase time model has better prediction hit rate than a basic model. 3. By using the multiplicative model, our multi-category model can estimate the behavior of product transitions between two consecutive purchases.