Prediction of Consumer Purchases in a Session on an EC Site Considering the Variety of Past Browsing

Data obtained from online shops include not only consumer purchase information but also data on non-purchase behaviors. This enables the construction of models to predict consumer purchases during a session. Behaviors such as focusing on a single product or repeatedly visiting the site for product consideration are thought to influence in-session purchases. Specifically, it is believed that consumers who show a keen interest in a particular product and spend time examining it are more likely to make a purchase compared to those who evaluate and compare various attributes of products in a specific category. Additionally, even if consumers spend a significant amount of time viewing a particular product, they might do so solely for information gathering. Furthermore, consumers who have repeatedly visited the site or have made multiple purchases in the past are likely to be more familiar with purchasing on the site compared to first-time visitors, making them more inclined to make a purchase. In summary, such factors can be crucial in accurately predicting consumer purchase intentions. However, these aspects of consumer purchasing behavior have not been reflected in traditional models. Therefore, this research aims to construct a model that accounts for the diversity of consumer purchasing behaviors and visit histories, improving the accuracy of predicting purchases during a session beyond what traditional models provide. The results are intended to assist in decision-making for marketing strategies.