A Machine Learning‑Based Business Analytics Framework for Customer Churn Prediction in Nigerian Retail SMEs

This independent research paper develops and evaluates a machine learning‑based business analytics framework for predicting customer churn in Nigerian retail SMEs. Using a synthetic but realistic dataset of 1,000 customer records, the study applies logistic regression, random forest, and gradient boosting models to identify behavioural and transactional predictors of churn. Gradient boosting achieved the highest performance with an accuracy of 88.9%, precision of 0.87, recall of 0.90, F1‑score of 0.88, and ROC‑AUC of 0.93. Engagement score, purchase frequency, tenure, and recency emerged as the most influential predictors. The study provides a practical, low‑cost analytics framework that SMEs can adopt to transition from reactive to proactive customer retention. This work forms part of the author’s ongoing portfolio in data science, business analytics, and applied machine learning, and has undergone independent peer review. It is not an official university thesis.