A national telecom operator · XGBoost + Random Forest · Customer Segmentation · ARPU Optimization
Increasing churn in broadband prepaid services — customers migrating to competitors at growing rates.
No predictive mechanism to identify customers likely to churn before they actually leave the network.
No data-driven model to identify optimal moments for offering upgrades and increasing average revenue per user.
One-size-fits-all retention offers with no segmentation by behavior (Power/Mild/Low) or location (urban/rural).
XGBoost + Random Forest on usage, recharge, network quality, and package data to predict at-risk customers.
Identifies optimal upgrade moments per customer segment — boosting conversion rates and ARPU.
Power / Mild / Low usage profiles with an urban-rural split for localized, personalized interventions.
Upgrade/downgrade analysis, refined usage segmentation, and dynamic behavioral modeling for rural/urban populations.
Churn predictions for targeted retention.
Upsell model drives revenue growth per user.
Engagement via behavior + location segmentation.
ML models on usage, recharge, network and QoS data.