This document discusses using machine learning models to predict customer churn for a telco company and provide targeted recommendations to reduce churn. The data was explored and customers were segmented into 5 clusters based on usage. Decision tree and random forest models were used to determine why customers churn and which customers churn for each cluster. Targeted recommendations were then provided for different customer categories within each cluster, such as offering free international minutes, discounted calling rates, or prioritized customer service. The results showed the potential to avert $555,700 in annual revenue loss through reducing customer churn.