🧠 Customer Churn Prediction Project

📋 Overview

This machine learning project predicts customer churn using a Random Forest Classifier.
The goal is to identify customers who are likely to leave based on behavioral and demographic data.

---

📁 Project Structure

churn_prediction/
│
├── data/                # Dataset files (ignored in repo)
│   └── generate_data.py  # Script to generate sample data
│
├── models/              # Trained model files (ignored in repo)
│
├── notebooks/           # Jupyter notebooks for analysis
│   └── churn_analysis.ipynb
│
├── src/                 # Source code modules
│   ├── data_preprocessing.py
│   ├── model_training.py
│   ├── utils.py
│   └── __init__.py
│
├── main.py              # Main execution script
├── predictions.csv      # Sample predictions output
├── requirements.txt     # Python dependencies
└── README.md

---

⚙️ Installation

Make sure you have Python 3.8+ installed.
Then install all dependencies:

pip install -r requirements.txt

---

🚀 Usage

To run the full prediction pipeline:

python main.py

Or train and test the model manually:

python src/model_training.py
python main.py   # Generates predictions.csv

---

📄 Files Description

File	Description
data_preprocessing.py	Handles data cleaning and encoding
model_training.py	Trains and evaluates the Random Forest model
utils.py	Utility functions for predictions
main.py	Runs the full churn prediction pipeline

---

📊 Results

Model: RandomForestClassifier (n_estimators=100, max_depth=15)

Evaluation on Test Set:

• Accuracy: 0.912
• Precision: 0.89
• Recall: 0.87
• F1-score: 0.88
• ROC-AUC: 0.94

You can reproduce these results using:

python src/model_training.py

The predictions are saved to:

predictions.csv

---

📦 Output

The output file predictions.csv contains churn probabilities for each customer.
Example:

Customer_ID	Churn_Probability
1001	0.12
1002	0.84
1003	0.45

---

🧩 Data & Model Files

The dataset (customer_churn.csv) and trained model (churn_model.pkl) are not included in this repository to keep it lightweight.

You can generate them yourself by following these steps:

1. Generate synthetic data

   python data/generate_data.py

   This creates data/customer_churn.csv.

2. Train the model

   python src/model_training.py

   This saves the trained model as models/churn_model.pkl.

3. Run predictions

   python main.py

   This creates a predictions.csv file with churn probabilities.

---

🧾 License

This project is open-source and available under the MIT License.

---

👨‍💻 Author

Sarim Shah — GitHub: @ssarimm