README.md

metadata

title: Customer Churn Predictor
emoji: 📊
colorFrom: blue
colorTo: indigo
sdk: gradio
sdk_version: 5.29.1
app_file: app.py
pinned: false

🧠 Customer Churn Predictor – Gradio App

A machine learning web app to predict customer churn using the Telco Customer Churn dataset. Trained using XGBoost and deployed with Gradio on Hugging Face Spaces.

🚀 Demo

Enter customer details to predict the likelihood of churn. The model analyzes usage behavior, contract type, billing preferences, and more to estimate the risk of a customer leaving.

📂 How It Works

• Preprocessed the Telco dataset (cleaning, encoding, scaling)
• Trained multiple models: Logistic Regression, Random Forest, XGBoost
• Tuned hyperparameters for best performance (XGBoost selected)
• Saved model and required metadata with joblib
• Built a Gradio UI for real-time inference
• Deployed to Hugging Face Spaces for public use

📈 Example Inputs

Feature	Type	Example Value
SeniorCitizen	Binary	0
Tenure	Numeric	12
MonthlyCharges	Numeric	79.5
TotalCharges	Numeric	945.3
Contract	Categorical	Month-to-month
InternetService	Categorical	Fiber optic
PaymentMethod	Categorical	Electronic check

🧪 Model Info

• Algorithm: XGBoost Classifier
• Accuracy: ~84%
• Preprocessing: One-hot encoding, StandardScaler

📦 Dependencies

See requirements.txt in the repo.

🙋 Author

Abhishek Singh
Research Analyst & ML Enthusiast
GitHub | LinkedIn

---

🚧 Note

This app is for educational/demo purposes using open data from Kaggle.