📉 Customer Churn Modelling using Artificial Neural Network (ANN)

This project predicts customer churn using an Artificial Neural Network (ANN) and provides an interactive web interface built with Streamlit. Users can input customer details and instantly get a churn prediction.

🔗 Live App: 👉 https://churnmodellingusingann-jxyqbzwvxtc9ndrnwtqbo9.streamlit.app/

---

🚀 Project Overview

Customer churn prediction helps businesses identify customers who are likely to leave. In this project:

• An ANN model is trained on banking customer data
• The trained model predicts whether a customer will churn or not
• A Streamlit web app is used for real-time predictions
• Preprocessing objects and trained models are saved and reused

---

🌟 Key Features

• ANN-based churn prediction
• Interactive Streamlit UI
• Real-time predictions
• Data preprocessing (encoding & scaling)
• Saved model and encoders (.pkl, .h5, .keras)
• User-friendly web deployment

---

🛠️ Technologies Used

• Python
• NumPy
• Pandas
• Scikit-learn
• TensorFlow / Keras
• Streamlit
• Jupyter Notebook

---

📂 Project Structure

churn_modelling_usingANN/
│
├── BankChurners.csv              # Dataset
├── Churn_Modelling.csv           # Dataset
│
├── app.py                        # Streamlit application
├── experiments.ipynb             # Model training & experiments
├── predictions.ipynb             # Prediction testing
│
├── model.h5                      # Trained ANN model
├── model.keras                   # Keras model format
│
├── scaler.pkl                    # StandardScaler
├── ohe_geo.pkl                   # One-Hot Encoder (Geography)
├── label_encoder_gender.pkl      # Label Encoder (Gender)
│
├── requirements.txt              # Dependencies
├── req.txt                       # Alternate requirements file
└── .devcontainer/                # Dev container setup

---

📦 Installation (Local Setup)

Clone the repository:

https://github.com/JahnaviSingh2005/churn_modelling_usingANN.git

Move into the project directory:

cd churn_modelling_usingANN

Install dependencies:

pip install -r requirements.txt

---

▶️ Run the Streamlit App Locally

streamlit run app.py

---

📊 Input Features

The model uses the following inputs:

• Credit Score
• Geography
• Gender
• Age
• Tenure
• Balance
• Number of Products
• Has Credit Card
• Is Active Member
• Estimated Salary

---

🎯 Output

• 0 → Customer will NOT churn
• 1 → Customer WILL churn

The result is displayed instantly on the Streamlit interface.

---

🌐 Live Deployment

The application is deployed using Streamlit Cloud.

🔗 Live App Link: https://churnmodellingusingann-jxyqbzwvxtc9ndrnwtqbo9.streamlit.app/

---

🔮 Future Enhancements

• Model performance comparison with ML algorithms
• Hyperparameter tuning
• Explainable AI (SHAP / LIME)
• Improved UI/UX
• Dataset expansion

---

👩‍💻 Author

Jahnavi Singh B.Tech Student | Machine Learning & AI Enthusiast

---

Just say the word 🚀