RetainX — Customer Churn Prediction for Subscription-Based Services

Team 7: Colorado Blue Spruce
Kushal Shankar | Ashmitha Appandaraju | Tej Sidda | Ronak Vadhaiya

📈 Project Overview

RetainX is a predictive analytics tool designed to help subscription-based businesses proactively identify customers likely to churn and implement targeted retention strategies.

---

📂 Project Structure

Folder	Contents
data/	Preprocessed Telco Customer Churn dataset
models/	Saved XGBoost best model for interpretation
notebooks/	Jupyter notebooks for each stage: EDA, preprocessing, modeling, interpretation
results/	Visuals and plots generated from the notebooks (PNG files)
exports/	PDF exports of each notebook and the presentation
report/	Final project report
team_testing_checklist.md	Documented testing done by each team member

---

✅ Workflow Summary

• Data Cleaning & EDA
• Preprocessing & Feature Engineering
• Baseline Models (Logistic Regression, Decision Tree)
• Advanced Modeling (Random Forest, XGBoost)
• SHAP Explainability
• Business Insights and Recommendations

---

📚 Dataset

Telco Customer Churn dataset (IBM Sample on Kaggle)

---

🛠 Technologies Used

• Python: Pandas, NumPy, Scikit-learn, XGBoost, SHAP, Matplotlib, Seaborn
• Jupyter Notebooks
• Joblib (for model persistence)

---

✅ Key Findings

• Customers with month-to-month contracts and high monthly charges are at the highest risk of churn.
• Electronic check payment method users are more prone to churn.
• Tenure is a strong retention factor.

---

📥 Reproducibility

git clone https://github.com/Kushal-Shankar-1/RetainX.git
cd RetainX_Final_Submission/
pip install -r requirements.txt
jupyter lab

Then run notebooks in order inside /notebooks.

---

✅ Conclusion

RetainX offers an end-to-end churn prediction and analysis workflow and provides actionable business recommendations for proactive retention strategies.