Machine Learning Zoomcamp - Capstone Project - Wine Quality

Description

This project is created as the capstome project of the Machine Learning Zoomcamp, and it will be peer reviewed and scored.

The objective is to build a model to predict the quality of the wine.

Dataset

This project uses the wine quality datasets available at the UCI website. The two datasets are related to red and white wine samples, from the north of Portugal.

System architecture and Technologies

The following figure depicts the system architecture and technologies used:

Setup

This project has as main pre-requisites:

• Python 3.10
• Docker 20.10.20
• Bentoml 1.0.8

How to run:

1. Clone this repo
2. Open a terminal within that folder
3. Run pip install pipenv
4. Run pipenv install, to install the dependencies from Pipfile
5. Run pipenv shell
6. Run python train.py
7. Run bentoml build
8. Run bentoml containerize wine
9. Run docker run -it --rm -p 3000:3000 wine:<tag> serve --production, where tag is obtained in the previous step.
10. Finally, run python test_predict.py to send API requests to test it, or use the Swagger-UI as shown in the following figure: