An enterprise-grade, end-to-end churn prediction and risk management system built with Streamlit and Scikit-Learn. This platform transforms raw telecom data into actionable business intelligence, helping retention teams prioritize high-value customers and optimize marketing spend.
- Real-time KPIs: Monitor Churn Rate, Total Revenue at Risk, and Average Monthly Charges.
- Revenue Impact: Instant visibility into potential monthly and total revenue loss.
- Segmentation: Churn distribution by Gender, Seniority, and Partner status.
- Feature Correlation: Interactive horizontal bar charts showing exactly what drives churn.
- Service Impact: Analysis of how support services (Online Security, Tech Support) influence retention.
- Driver Analysis: Deep dives into Tenure and Monthly Charges using distribution box plots.
- Multi-Model Training: Trains and evaluates Logistic Regression, Random Forest, and Gradient Boosting.
- Unified Pipeline: Intelligent preprocessing (Label Encoding + Standard Scaling).
- Performance Metrics: Full evaluation suite including ROC-AUC, F1-Score, and Precision-Recall curves.
- Input Form: Assess any individual customer by entering their demographics and services.
- Probability Scoring: Get an exact churn percentage.
- Actionable Advice: Dynamic retention recommendations based on risk levels (High/Medium/Low).
- Global Importance: See which features the model values most across the entire dataset.
- SHAP values: High-fidelity explanations for why the model makes specific predictions.
- Demographic Parity: Analyze if the model is biased against specific genders or age groups.
- Fairness Metrics: Automated disparity calculation to ensure ethical AI deployment.
- Cost-Benefit Analysis: Input your own business costs for false alarms vs. missed churn.
- Campaign Optimization: Automatically allocates a retention budget across risk segments.
- ROI Tracking: Projected Revenue Recovery and Net Benefit summary.
- Frontend: Streamlit (Data App Framework)
- Analysis: Pandas, NumPy
- Visualizations: Plotly Express, Plotly Graph Objects, Seaborn, Matplotlib
- Machine Learning: Scikit-Learn (Classification, Preprocessing, Metrics)
- Explainability: SHAP (SHapley Additive exPlanations)
- Deployment: Joblib (Model Serialization)
-
Clone the repository:
git clone https://github.com/Codehari04/Enterprise-Telecom-Risk-Intelligence.git cd "Enterprise Telecom Customer Risk Intelligence Platform"
-
Install Dependencies: Ensure you have Python 3.8+ installed.
pip install -r requirements.txt
-
Run the Application:
streamlit run app.py
Hariharan
- Email: hariharan22td0674@svcet.ac.in
- GitHub: Codehari04
This project is designed and developed for Data-Driven Decision Making. All rights reserved. © 2025 Telecom Analytics Enterprise.