An AI-powered web application built with Streamlit that predicts whether a telecom customer will churn (leave the company) or stay, based on their profile and service usage details.
screen-capture.17.webm
- Predicts Customer Churn Risk using a trained ML model.
- Two modes:
- 🧑 Single Customer Prediction
- 📄 Batch Prediction (CSV upload)
- Probability-based predictions with clear visualization.
- User-friendly Streamlit interface with modern sidebar design.
- Supports both categorical and numerical features.
- Open the app in your browser.
- Choose between Single Customer or Batch Prediction (CSV).
- For single prediction: Fill in customer details and click Predict Churn.
⚠️ Yes → Customer WILL churn.- ✅ No → Customer will NOT churn.
- For batch prediction: Upload a CSV file with customer data and view predictions.
The app is trained using the Telco Customer Churn Dataset from IBM Sample Data.
-
Target Classes:
Yes→ Customer will churnNo→ Customer will stay
-
Features:
- Demographics (Gender, SeniorCitizen, Partner, Dependents)
- Account Info (Tenure, Contract, Payment Method, Paperless Billing)
- Services (Phone, Internet, Online Security, Streaming, Tech Support, etc.)
- Charges (MonthlyCharges, TotalCharges)
- Python 3.9+
- Streamlit (Frontend Web App)
- Pandas & NumPy (Data Processing)
- Matplotlib & Seaborn (EDA & Visualization)
- Scikit-learn (Label Encoding, Model Training, Evaluation)
- XGBoost & Random Forest (Machine Learning Models)
- SMOTE (imblearn) (Handling Class Imbalance)
- Pickle (Model & Encoder Serialization)
Mirza Yasir Abdullah Baig
This project is for educational purposes only and should NOT be used for real-world business decisions without further validation.