Tutorial

Welcome, this doc provides a tutorial on how to get quickly started on the usage of Carol platform for Data Science, Machine Learning and Analytics projects. The chapters of this document covers both the theoretical part as well as an hands-on demo through a classic regression example: Boston House Pricing.

This documentation is also available at github, at this link, where it contains the source code for the examples we will be working on.

We start with an overview on carol components and how to find them on the navigation user interface (chapter 1), next we explore how to load your own data and build higher level data models over these records (chapters 2 and 3).

On chapters 4 and 5 we explore the construction of batch apps on top of Carol, explorying a very basic Jupyter based app to start, and then a robust task management using Luigi.

The chapters 6 and 7 are dedicated to examplify how to publicate your code as a service (API based on an online App) and with an additional layer of basic UI.

We hope this tutorial brings you up to speed and help you build powerfull tools on Carol!

• 1. Introduction
• 2. Copying data to Carol
• 3. Creating a data model
• 4. Training the model: a minimal batch app
• 5. Training the model: task orchestration with Luigi
• 6. Online app: Serving requests through a REST API
• 7. Online app: Building a simple GUI with Streamlit
• 8. Appendix