隨著網路商店的崛起，瀏覽資料成為商家積極想要分析的部分，而零售商更 是我們生活最息息相關的產業，現今人們的購買範圍已經不僅限於線下，本研究 的目的即是探討如何藉由零售商的線上瀏覽資料解析出更多背後意涵，並更深入 地洞悉和理解每次客戶來訪時的購物行為和意圖，從而使電商能夠根據每次來訪 的需求為其提供量身定制而使其滿意的服務。 此研究提出一種結合監督式與非監督式的商業分析方法，先將每筆資料整理 成以工作階段(Session)為單位，再以 K-means 為主做分群，將資料分成五群，再 根據分群狀況使用決策樹去找出特徵值且命名。 將理論與實務結合，我們擬合此五群與客戶旅程(Customer Journey)之理論， 第一群至第五群皆符合察覺(Awareness)階段、思考(Consideration)階段、購買 (Purchase)階段，客戶旅程裡的前三階段，隨後也給予根據客戶旅程的行銷策略建 議。此外，在第四章文末更加入預測分析，預測顧客在工作階段內是否進入購物 車，藉此更提升本研究對瀏覽資料分析的全面性。;Online consumer browsing behavior has been widely researched by many companies and researchers. The retail industry is also the most common industry in the world. Nowadays, people not only purchase offline but also purchase online and even more often than before. This study extends the analysis of current research on finding more insights behind browsing data. The goal is also to delve deeper into and comprehend the customers’ shopping behavior and intentions per visit and, thus, enable retailers to provide customers’ satisfactory services tailored to their requirements per visit. We demonstrate the utility of our approach by applying it to a real case of a retailer. We propose a business analytics approach that utilizes integrated supervised and unsupervised learning. Firstly, we do the session-based data preprocessing, after which we use K-means to find 5 clusters and name them after their rules and patterns by decision tree. On the integration of theory and practice, we map these 5 clusters with customer journey. The results show group 1 to group 5 just fit “Awareness”, “Consideration”, “Purchase” three phases. We also propose some marketing strategies around the customer journey. In addition, we use supervised learning to predict whether or not a customer in a session will go to the Shopping Cart, which makes this study more comprehensive.
