Table of contents

• General info
• Architecture
• Setup
• Discussion and conclusions

General info

This repository is made by data scientists for data scientits. Its aim is to reduce time searching for installers and libraries when starting a new project. It also considers the necessity to orchestrate the ML pipelines with Airflow and to track and register the models with MLFlow.

Architecture

NOTE:

• The python and airflow container is based on puckel with some modifications due to specific requirements of snap7 library.
• There is a fifth container, hidden in the architecture diagram, to wait for dependencies. It is courtesy of dadarek.

Setup

To run this project, clone the repo and run the next command in the project folder:

docker-compose -f docker-compose-skeleton.yml up

NOTE: it is necessary to have Docker and Docker Compose installed in your host machine.

Discussion and conclusions

Any insight or suggestion in order to correct bugs or improve the architecture is welcome. The goal is to facilitate the job to the data scientits at the beginning of the projects.