Recommendations to Increase the Customer Interaction of E-commerce Applications with Web Usage Mining

Abstract

Web mining uses data mining methods to extract knowledge from web applications. It is used in e-commerce to track client browsing behavior. The issue with e-commerce is that we only know about our customers once they place an order. The primary goal of this endeavor is to identify a viable option to continue operating a profitable online store by better understanding customers. The research project aims to identify customer preferences and purchase behaviors so that improvements can be made to e-commerce platforms based on these findings. Web usage mining enables the seller to monitor, investigate, and identify patterns from compiled data to create a fundamental statistical foundation for decision-making. To properly use web usage mining, it is necessary to collect qualitative visitor data, which enables researchers to determine whether a visitor has viewed a product repeatedly, added it to their wishlist before making a purchase, or bought it during a particular season, etc. A limited number of publications were found in this domain, and most of the work was done with limited data like user clicks, navigation paths, etc. In this research, event listeners have been added to the e-commerce application to capture user actions and behavior towards a specific product on the application. Four classification algorithms experimented with the event data and identified the most effective algorithm to develop the prediction model. Users' purchasing patterns and buying behaviors were analyzed and identified using the model developed with the Random Forest classification algorithm. Recommendations for the e-commerce applications were developed according to the identified user behavior and purchasing patterns. This strategy will lead the ecommerce industry to a profitable economic point by increasing the effectiveness of the application.