About

This project builds a production-ready customer churn prediction system using real hotel review data. The workflow transforms review scores and text length into churn risk signals, trains a validated logistic regression model, and evaluates performance using accuracy, precision, recall, and ROC-AUC. The system is deployed as an automated, scheduled pipeline in Zerve, demonstrating how predictive analytics can be operationalized to proactively identify at-risk customers and support retention strategies.