Skip to content

README.md

Latest commit

 

History

History
163 lines (105 loc) · 4.13 KB

README.md

File metadata and controls

163 lines (105 loc) · 4.13 KB

📊 Customer Churn Prediction App

Python Framework ML Package Manager License Status

End-to-end Machine Learning project for predicting customer churn.

Includes:

  • Data preprocessing

  • Model training (Random Forest)

  • Pickle model generation

  • Deployment using FastAPI, Flask, Streamlit, Gradio

  • HTML + CSS frontend

🚀 SETUP INSTRUCTIONS (STEP-BY-STEP)

1️⃣ Fork the repository (from GitHub UI)

2️⃣ Clone your fork

git clone https://github.com/YOUR-USERNAME/YOUR-REPO-NAME.git

cd YOUR-REPO-NAME

3️⃣ Install uv (if not installed)

Mac / Linux

curl -Ls https://astral.sh/uv/install.sh | sh

Windows (PowerShell)

irm https://astral.sh/uv/install.ps1 | iex

Verify installation

uv --version

4️⃣ Create Virtual Environment

uv venv

Activate virtual environment

Windows

.venv\Scripts\activate

Mac / Linux

source .venv/bin/activate

5️⃣ Install Project Dependencies

uv sync

6️⃣ Train Model (Generate Pickle File)

Make sure dataset exists:

data/churn.csv

python train_model.py

This creates:

models/churn_model.pkl

🌐 RUN APPLICATIONS

▶️ Run Flask App

python deployments/flask/flask_app.py

Open in browser:

http://127.0.0.1:5000/

▶️ Run FastAPI App

uvicorn deployments.fastapi.fastapi_app:app --reload

Open:

http://127.0.0.1:8000/

Swagger Docs:

http://127.0.0.1:8000/docs

▶️ Run Streamlit App

streamlit run deployments/streamlit_app.py

▶️ Run Gradio App

python deployments/gradio_app.py

📂 PROJECT STRUCTURE

.
├── data/
│   └── churn.csv
├── models/
│   └── churn_model.pkl
├── deployments/
│   ├── flask/
│   ├── fastapi/
│   ├── streamlit_app.py
│   └── gradio_app.py
├── train_model.py
├── pyproject.toml
└── README.md

📌 IMPORTANT NOTES

  • Train the model before running any app.

  • Entire preprocessing + model is saved inside pickle.

  • No manual encoding required in deployment.

  • SMOTE (if used) is only for training, never for production API.

🎯 DONE

You now have a fully working ML project with multiple deployment options.
⚡