Identifying Customer Churn for E-Commerce Using Ensemble Learning Method

Organizations in the E-Commerce (E-Comm) sector, particularly those belonging to the business-to-consumer category are in strong fight for their survival, attempting to acquire accessibility with their competitors' client groups while preventing existing customers from leaving. The expense of obtaining novel customers is increasing as additional competitors enter the market with large initial investments as well as cutting-edge penetration methods, making client retention critical for these businesses. The most appropriate course for proceeding in this situation is to recognize potential churning clients and avert churn using temporary maintenance tactics. Understanding wherefore the consumer opted for the route is also necessary in order to implement specific win-back methods. The e-commerce corporation keeps every client's records on file comprising searches done, purchase fascinated, purchase frequency, reviews left, feedback provided, and additional data. The combination of Machine Learning (ML) with clickstream data analysis has transformed E-Comm platforms. In the E-Comm context, this study provides a comprehensive examination of the ML usage to estimate client purchase decisions, identify engagement variables, and improve user interfaces. The data used in this research employs ML to assess client behavior in E-Comm was obtained from a variety of sources among 2004 and September 2023. These models provide as the foundational learners in an ensemble structure. A weighted-average ensemble approach integrates them, yielding an impressive 99.84% accuracy on a reserved test dataset. This strategy is more promising than previous approaches for detecting client attrition. In marketing, company relationships are critical for transactions over the internet.