A User Behavior Analysis Method Based on Big Data

Abstract

With the developing progress of the Internet and the pandemic deterioration, more people decide to shop online. The online eCommerce business is quickly expanding. Understanding users’ purchase behavior is a key component to growing sales and improving the website’s design and flow. We use User Behavior Data from Tianchi of Alibaba Cloud. The data were randomly selected about one million users who have behaviors of page view, purchase, shopping to cart, and adding to favorite during November 25 to December 27, 2017. There are a total of ten million records. We used four dimensions to approach this problem: time, item, conversion rate from Funnel Analysis, and valuable customers from RFM Model. From the time dimension, users are more likely to visit pages during the weekends from 10 am to 3 pm. From the item dimension, the top ten items were purchased over one hundred times during this time frame. The number is outstanding back in 2017. For the conversion rate from Funnel Analysis, adding items into the cart has a higher conversion rate compared to marking items as favorites. For the customer dimension, we used an RFM (recency, frequency, monetary) model to identify valuable users. We can advise Taobao to improve the favorite item remainder process, increase the exposure of hot selling items during peak hours and week of the day, and offer more promotions/recommendations to valuable customers.