Evaluating Machine Learning Techniques for Predicting Customer Churn in E-Commerce: A Comparative Analysis

Abstract

This study intends to demonstrate different machine-learning techniques for business intelligence applications. These techniques, identified as clustering, decision trees, naive Bayes, support vector machines (SVM), and logistic regression, provide their utility for e-commerce customer churn forecasts, as one of the highlighted methodological procedures, thematic analysis was used against prevailing literature to recognise strengths, weaknesses and probable implementations of every technique, specifically in business intelligence. Integrating machine learning techniques for predicting customer churn rate can be considered a contributory aspect of this study. Identification of differences can be useful for e-commerce owners to avoid undesired issues in business operations. These themes are meant to fulfil the aim of this research, to contrast different machine learning methods. Further, the study identified the significance of technique choice as dependent on e-commerce dataset aspects, exchanges within interpretable capability, precision, and computational efficacy. The overall findings, divided into specific themes, contribute to the identification of the benefits of each technique. The findings, through comparison, added specific inference to the flexibility, simplification and suitability of decision trees for categorical data management. The research study guides e-commerce in implementing machine learning as a practical implementation for diminishing business customer churn. Additionally, the research provides a prospective scope for future studies for observational appraisal regarding the discussed techniques applying realistic e-commerce data to validate their efficiency.