🔥 ANN-Classification-Customer-Churn

A lightweight Streamlit app that predicts whether a customer is likely to churn based on demographic and service-related input features. Powered by a trained TensorFlow model.

![App Screenshot]

🚀 Live Demo

👉 Try the live app on Hugging Face Spaces:
[https://huggingface.co/spaces/KaustavModak/ann-classification-customer-churn]

---

🧠 Model Overview

• Built with TensorFlow 2.12.0
• Preprocessing using:
  • LabelEncoder (Gender)
  • OneHotEncoder (Geography)
  • StandardScaler (Numerical features)
• Input features include:
  • Age, Gender, Geography
  • Credit Score, Balance, Tenure
  • Number of Products, Has Credit Card, Is Active Member

---

📂 Project Structure

.
├── app.py                      # Streamlit frontend
├── model.h5                   # Trained Keras model
├── label_encoder_gender.pkl   # Gender encoder
├── onehot_encoder_geo.pkl     # Geography encoder
├── scaler.pkl                 # Scaler for numerical features
├── requirements.txt           # Python dependencies
├── runtime.txt                # Python version pin
├── README.md                  # This file

---

⚙️ Setup Instructions

🔧 Run locally

# Clone the repository
git clone https://github.com/yourusername/churn-predictor.git
cd churn-predictor

# Create and activate virtual environment
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Run Streamlit app
streamlit run app.py

📊 Example Prediction

Feature	Value
Age	35
Gender	Male
Geography	France
Credit Score	720
Tenure	5 years
Balance	50000
Products	2
Has Credit Card	Yes
Is Active Member	No

✅ Predicted: Likely to Stay

---

📜 License

This project is licensed under the MIT License.

---