Telecom Customer Churn Analysis
This project focuses on reducing customer churn in the telecom industry using predictive analytics and business intelligence. By applying Logistic Regression and Random Forest models in R and designing an interactive Power BI dashboard, I identified the top drivers of churn and built actionable recommendations.
Key findings show that month-to-month contracts (23.5%), short-tenure customers (<12 months), and those with high monthly charges are most likely to churn. Lack of tech support and security features further increases risk.
The analysis compared two models: Random Forest (87.5% accuracy) and Logistic Regression (80.2%). While Random Forest performed better, Logistic Regression was chosen for its interpretability and ease of communication with business stakeholders.
The Power BI dashboard visualizes churn patterns, model comparisons, and feature impacts, making the insights practical for managers. Based on the study, I recommended incentivizing long-term contracts, bundling support services, promoting auto-pay options, and improving early onboarding.
This end-to-end solution demonstrates how data science + BI tools can transform churn management into a strategic advantage.