Customer Churn Prediction

Predict customer churn using a deep learning model built with TensorFlow and Keras. This project provides an end-to-end pipeline for analyzing customer attributes, engineering features, training a model, and making predictions—all interactively demonstrated within a Jupyter notebook.

---

🚀 Project Overview

Customer churn prediction aims to identify customers who are likely to leave (churn) a business. By anticipating churn, companies can proactively engage at-risk users and improve retention.

Key Steps:

• Data loading and inspection
• Feature engineering (including one-hot encoding)
• Data scaling (StandardScaler)
• Model building with Keras (Dense, Dropout layers)
• Training and evaluation (metrics, confusion matrix, ROC curve)

Interactive Demo:
See the Jupyter notebook: Customer_Churn_Prediction.ipynb

---

📊 Sample Data

The notebook uses the Churn_Modelling.csv dataset.
Example features:

• CreditScore, Geography, Gender, Age, Tenure, Balance
• NumOfProducts, HasCrCard, IsActiveMember, EstimatedSalary
• Target: Exited (1 = churn, 0 = retained)

---

🛠️ Technologies & Libraries

• Language: Python 3.x
• Data Manipulation: numpy, pandas
• Visualization: seaborn, matplotlib
• Deep Learning: tensorflow, keras
  • Model: Sequential, Dense, Dropout
  • Optimization: Adam
  • Callbacks: EarlyStopping

---

📓 Usage

1. Installation

pip install numpy pandas seaborn matplotlib tensorflow

2. Running the Notebook

1. Download Customer_Churn_Prediction.ipynb and Churn_Modelling.csv to the same directory.

2. Launch Jupyter:

   jupyter notebook
   # or
   jupyter lab

3. Open the notebook and run cells sequentially.

3. Contributing

• Fork this repo
• Create a new branch (git checkout -b feature-branch)
• Make your changes
• Commit and push
• Open a pull request

---

🌟 Features

• End-to-end pipeline: Data cleaning → Feature engineering → Deep learning.
• Interactive visualizations: Data exploration and evaluation plots.
• Easy to extend: Modular code, well-commented.
• No missing values: Data integrity checks included.
• Customizable architecture: Quickly tune layers, activations, and callbacks.

---

🙏 Acknowledgments

• Dataset: Kaggle - Churn Modelling Dataset
• TensorFlow & Keras documentation
• Community tutorials on customer churn prediction

---

🤔 FAQ

Q: Can I use my own data?
A: Yes! Just format your CSV file with similar columns and update the data loading cell in the notebook.

Q: How do I interpret the results?
A: The notebook explains model predictions with evaluation metrics and plots.

---

Feel free to open issues or contribute improvements!