This document discusses the development of a predictive churn model for Segment 9 subscribers in REC base. It defines churn, outlines the modeling process including creating an aggregated billing table (ABT) from multiple data sources, selecting important variables, and iterating the model 30-40 times. Key variables that impact churn are identified, such as average recharge amount and number of days since last recharge. The final section notes a model performance slide was included but its contents are not described.