網路消費者行為之網站造訪期間對購買期間的影響性－以訂購機票網站為例

Abstract

摘要 如何瞭解顧客的偏好結構，並準確預測顧客未來消費行為，一直是行銷人員孜孜不倦的方向。由於科技日益進步，龐大的資訊與強大的計算工具擴大了行銷人員瞭解消費者的能力。然而在消費者行為存在異質性的前提下，使用傳統的統計方法來描述消費者行為，需面臨以下抉擇：以全體顧客資料推估總體平均行為，卻忽略個別顧客之間的異質性；若僅依個別顧客資料推估個別顧客行為，則往往因資料量不足導致估計不具效率性的問題。 由於消費者行為存在異質性，行銷人員面對的並不是均質的大眾市場，因此行銷人員應該強調與顧客建立長期而穩定關係之關係行銷，以及考慮顧客異質性的一對一行銷。本研究主要目的是將層級貝氏理論運用於顧客網路購買期間之模型，進而達成估計顧客個人價值之目的。 本研究取樣自comScore資料庫，透過expedia旅遊網站之實際造訪與購買交易資料的分析，深入探討顧客的網路購買行為，以作為顧客價值分析的依據，希望能藉此提供企業進行資料庫行銷，並以此作為與顧客一對一溝通之基礎。再者，以層級貝氏理論所建立之顧客網路購買行為的預測模型，做為顧客分群的依據，比較不同顧客群之網路購買行為的差異。

Abstract How to understand the customer's preference structure and to predict future consumer behavior customer has been the tireless direction of marketing staff. Because of technology, huge information and computing power to expand the ability of marketers to understand consumers. However, the heterogeneity of consumer behavior under the premise that if we continued using the traditional statistical methods to describe consumer behavior, often face the following choice: to estimate the overall average of all customer information and behavior, but ignores the heterogeneity among individual customers; If only according to individual customer data to estimate individual customer behavior is often caused by insufficient information efficiency is estimated non-issue. Therefore, marketing staff are facing is no longer homogeneous mass market, but one by one with individual differences in customers, and customers should be emphasized to establish the relationship between long-term and stable relationship marketing, and customer heterogeneity into account a pair of become a mainstream marketing. The main purpose of this study is to use Bayes theory to customers during Internet shopping model, then customer personal worth is estimated to achieve the purpose. This study from comScore database through expedia travel site visit and purchase the actual trade data analysis, in-depth online buying behavior of customers, customer value analysis as a basis to provide database marketing as a business use communicate one to one basis with customers. Furthermore, in order to level the Bayesian theory to establish the customer network buying behavior prediction model, as the basis for grouping customers to compare different groups of customers online buying behavior differences.