ChurnPrediction – Customer Retention Classifier

Project Objective: Predict customer churn in a telecom company using classification algorithms to help reduce customer loss and guide retention strategies.

Technologies Used: Pandas, Scikit-learn, Matplotlib, Seaborn, XGBoost

Techniques Applied:

• Exploratory Data Analysis (EDA) and feature correlation.
• Handling missing values and encoding categorical variables.
• Balancing imbalanced data with SMOTE.
• Model training and tuning with cross-validation.

Results: The XGBoost classifier reached an accuracy of 0.83 with strong precision and recall, identifying key indicators of churn such as customer service calls, contract type, and tenure.

Skills Applied: Feature engineering, model selection, evaluation metrics (accuracy, precision, recall, F1-score), and deployment readiness.

Conclusions: The model can assist telecom companies in targeting at-risk customers with tailored retention strategies, potentially improving customer lifetime value.