A Data Mining Approach to Predict E-Commerce Customer Behaviour

Watsons, established in 1841, is one of the world’s leading beauty and personal care industry with more than 6300 stores in 11 different markets. In addition to 280 Watsons’s stores, online shopping is also an alternative for Turkish customers.

Usage of online tools and technologies for many different purposes has increased by 21st-century companies thanks to the great accessibility of the Internet. Therefore, due to current trends, many customers prefer online shopping. Among them, some of them are adding the products to their market basket but unfortunately; they are leaving the website without a purchase. This causes an important problem for most of the e-commerce retailers. In this context, research in e-commerce needs to determine the reasons that encourage them to buy as much as they need to understand the behaviours of consumers. However, it is difficult and complex to identify customer behaviour and the factors that encourage purchasing. Data mining techniques have become the focus of such analyses. This project mainly focuses on data classification approach to predict customer’s behaviour during the website visits of Watsons e-commerce channel.

Based on the problem to predict e-commerce customers’ behaviour at the customers’ first entrance to the Watsons’s website, data classification approach is studied. The collected data from Watsons’ e-commerce site are analysed in the WEKA by applying different data classification algorithms. Two distinct data sets are determined; while dataset 1 includes all attributes, dataset 2 is created by considering the results of attribute selection algorithms. Both dataset 1 and dataset 2 are examined separately in WEKA software by using seven classifier. The data classification algorithm with the highest accuracy is selected. Some rules and trees are suggested to the company to predict future customer behaviours. To encourage the customers who leave the website of Watsons, rules are interpreted for suggestion.