PredictingWineFeatures

This project predicts the best features of wines that relate to wine quality. This MLOps pipeline uses a self-managed Digital Ocean Github action-runner, Data Version Control to manage data changes across pipeline stages, and Continuous Machine Learning (CML) to support continual integration metrics analysis of models across stages, experiments, and branches.

Requirements

• On windows, WSL2, install using the WSL2 instructions, I've got the Ubuntu WSL2 environment installed.
• Python3.x and Pip3 are installed, I have a note in here to alias pip to pip3.
• Docker
• VS Code

Dataset

Modelling a Kaggle dataset of red wine properties and quality ratings.

Original Project

Original project https://github.com/andronovhopf/wine

Configure python's VirtualEnv and VSCode

Edit this to the end of ~/.bashrc, save.

alias pip="pip3"

And run these commands to install virtualenv on WSL2

source ~/.bashrc
pip install virtualenv

Validate virtualenv is installed

virtualenv --version
mkdir ~/.virtualenvs
virtualenv ~/.virtualenvs/PredictWindFeatures
source ~/.virtualenvs/PredictWindFeatures/bin/activate

Open VSCode, open the terminal in vscode and run this

python3 -m virtualenv .virtualenv

A dialog window will appear in the bottom right corner of VSCode saying the new virtual folder is detected, press "Yes" to use it. VS Code, install the extension, SonarLint, it will require JRE, let it download and install it. SonarLint will ask to restart VSCode, click "Restart Now".

Set up Python Requirements and Dependencies

Now you can run pip installs and they are only installed for your project.

Install from a requirements.txt document with:

pip install -r requirements.txt 

You can also run "pip freeze" to output the requirements for your python project or to pipe into a requirements.txt to freeze dependency versions.

pip freeze