utanvir/telco_churn

📉 Telco Churn Prediction Dashboard

End-to-end machine learning pipeline for predicting customer churn using the Telco dataset. Includes preprocessing, model training, evaluation, and deployment via a Streamlit dashboard.

🎥 Demo

[image]

🔧 Features

• —⚡ Modular pipeline architecture (clean, reusable code)
• —🧹 Preprocessing & feature engineering:
• —Missing value handling
• —Scaling, binning, one-hot encoding
• —Polynomial & interaction features
• —Feature selection with SelectKBest
• —🤖 Model training with hyperparameter tuning:
• —LogisticRegression
• —RandomForest
• —GradientBoostingClassifer
• —AdaBoostClassifier
• —SVC
• —KNeighboursClassifier
• —DecisionTreeClassifier
• —GuassianNB
• —MLPClassifier
• —XGBoost
• —📊 Evaluation metrics: Accuracy, Precision, Recall, F1, AUC-ROC
• —🎛️ Interactive Streamlit dashboard:
• —Exploratory Data Analysis (EDA)
• —Model evaluation & visualization (Confusion Matrix, ROC Curve)
• —Real-time churn predictions
• —CSV export of predictions

📁 Project Structure

churn-prediction-2/
├── artifact/                     # Stores intermediate artifacts
├── Chun_prediction.egg-info/     # Metadata for packaging
├── logs/                         # Log files
├── ml_venv/                      # Virtual environment (should be gitignored)
├── notebook/                     # Jupyter notebooks
│   ├── data/
│   │   └── telco_churn.csv       # Dataset
│   ├── eda.ipynb                 # Exploratory Data Analysis
│   └── MODEL TRAINING.ipynb      # Model training notebook
├── src/                          # Source code
│   ├── components/               # Core components
│   │   ├── data_ingestion.py
│   │   ├── data_transformation.py
│   │   ├── model_trainer.py
│   │   └── __init__.py
│   ├── pipeline/                 # Training & prediction pipelines
│   │   ├── train_pipeline.py
│   │   ├── predict_pipeline.py
│   │   └── __init__.py
│   ├── exception.py              # Custom exception handling
│   ├── logger.py                 # Logging utility
│   ├── utils.py                  # Helper functions
│   └── __init__.py
├── templates/                    # Flask HTML templates
│   ├── home.html
│   └── index.html
├── .gitignore
├── .python-version
├── app.py                        # Flask entrypoint
├── streamlit_app.py              # Streamlit dashboard
├── requirements.txt              # Dependencies
├── setup.py                      # Setup for packaging
└── README.md                     # Project documentation

🚀 How to Run

1. Clone the Repository

git clone https://github.com/your-username/churn-prediction_v2.git
cd churn-prediction_v2

2. Install Dependencies

pip install -r requirements.txt

3. Run Training Pipeline

python -m src.pipeline.train_pipeline

4. Launch Streamlit Dashboard

streamlit run streamlit_app.py

📦 Dependencies

• —pandas
• —matplotlib
• —seaborn
• —numpy
• —scikit-learn==1.6.1
• —imblearn
• —xgboost
• —dill
• —Flask
• —streamlit
• —(See requirements.txt for full details)

📊 Dataset

This project uses the Telco Customer Churn dataset available at: Kaggle Telco Churn Dataset (https://www.kaggle.com/datasets/blastchar/telco-customer-churn)

📌 License

MIT - You’re free to use, modify, and share it.

💡 Future Improvements

• —☁️ Cloud deployment (Hugging Face, AWS, Heroku)
• —🔎 AutoML (Optuna / RandomizedSearch)
• —📂 Handle unseen schema in uploaded CSVs
• —📈 Real-time monitoring dashboard with Streamlit metrics

🤝 Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

🔗 Connect with me: LinkedIn | Instagram