Cost-Effectiveness Analysis of Customer Churn and High-Value Customer Identification Using Random Forest on Online Retail Data

This study evaluates the cost-effectiveness of using the Random Forest algorithm to predict customer churn and identify high-value customers within online retail transactions. Leveraging the Online Retail Dataset from the UCI Machine Learning Repository, the research assesses key customer behaviors, the economic impact of predictive analytics, and return on investment in machine learning applications for customer retention. The study investigates the most influential factors in identifying high-value customers and compares the performance and cost-efficiency of Random Forest against other machine learning models. Adopting data-drive, decision making approach, findings will provide actionable insights for businesses to optimize customer engagement strategies while balancing predictive accuracy with financial feasibility, ultimately contributing to AI-assistant decision-making in e-commerce. This research fosters inclusive dialogue and shared leadership in data-driven business transformation.

Haiyan Bai; Fangxing Bai; Oluwaseun Farotimi; Comfort Omonkhodion