Iris Classification Pipeline, Streamlit App & FastAPI

This project demonstrates an end-to-end machine learning pipeline using the Iris dataset. It includes:

• A training script that builds an automated ML pipeline (with preprocessing, model training, evaluation, and saving).
• A Streamlit web application for model inference via a graphical interface.
• A FastAPI application to serve model predictions via REST API endpoints.
• A batch file that reads a .env file to initialize your virtual environment.

Project Structure

├── app.py             # Streamlit app for model inference
├── api.py             # FastAPI app for serving predictions via REST API
├── train_model.py     # Script to build, train, and save the ML pipeline
├── iris_pipeline.pkl  # Saved model pipeline (generated by train_model.py)
├── .env               # Environment variable file containing VENV path (to be created)
├── init_venv.bat      # Batch file to initialize the virtual environment
├── requirements.txt   # Python dependencies
└── README.md          # This file

Prerequisites

• Python 3.6+
• pip (Python package installer)
• Windows OS (for the provided batch file)

Setup & Operation Order

1. Clone the Repository

Clone this repository and navigate into its directory:

git clone https://github.com/mohiteamit/simple_pipeline_using_iris.git
cd simple_pipeline_using_iris

2. Create the .env File

Create a file named .env in the project root with the following content.
Make sure to replace <path-to-your-virtual-environment> with the actual path to your virtual environment's activation script:

VENV=<path-to-your-virtual-environment>\Scripts\Activate

3. Create/Activate the Virtual Environment

You have two options:

Option A: Using the Provided Batch File

1. With the .env file in place, run the batch file to initialize your virtual environment:

   init_venv.bat

Option B: Manually Activating Your Virtual Environment

If you prefer manual activation, run:

<path-to-your-virtual-environment>\Scripts\Activate

4. Install Dependencies

Once your virtual environment is active, install the required packages using requirements.txt:

pip install -r requirements.txt

5. Train the Model

Run the training script to load the Iris dataset, preprocess the data, train a RandomForest classifier (within a pipeline), evaluate its performance, and save the pipeline to disk:

python train_model.py

After running the script, a file named iris_pipeline.pkl will be generated.

6. Run the Streamlit App

Launch the Streamlit app to interact with the model via a graphical web interface:

streamlit run app.py

In your web browser, you can:

• Input iris feature values (sepal length, sepal width, petal length, petal width)
• Click the Predict button to view the predicted iris species

7. Run the FastAPI App

To serve predictions via a REST API, launch the FastAPI server with:

uvicorn api:app --reload

By default, the server will start at http://127.0.0.1:8000. You can access the interactive API documentation at http://127.0.0.1:8000/docs to test the /predict endpoint.

Additional Notes

• Automated Retraining:
  You may schedule the training script using tools like cron jobs, Windows Task Scheduler, or advanced scheduling tools such as Airflow if needed.

• Modularity:
  The training and inference stages are separated. You can update the model by re-running train_model.py without affecting the Streamlit or FastAPI interfaces.

• Customization:
  Feel free to extend the pipeline (e.g., add more preprocessing steps) or enhance the Streamlit app and FastAPI endpoints (e.g., add logging, error handling, or additional visualizations).

App screenshot

Contributing

Contributions and improvements are welcome. Please open an issue or submit a pull request for any suggestions or enhancements.

License

This project is licensed under the MIT License.

Contact

For questions or further information, please reach out to [[email protected]].