Book Image

Comet for Data Science

By : Angelica Lo Duca
Book Image

Comet for Data Science

By: Angelica Lo Duca

Overview of this book

This book provides concepts and practical use cases which can be used to quickly build, monitor, and optimize data science projects. Using Comet, you will learn how to manage almost every step of the data science process from data collection through to creating, deploying, and monitoring a machine learning model. The book starts by explaining the features of Comet, along with exploratory data analysis and model evaluation in Comet. You’ll see how Comet gives you the freedom to choose from a selection of programming languages, depending on which is best suited to your needs. Next, you will focus on workspaces, projects, experiments, and models. You will also learn how to build a narrative from your data, using the features provided by Comet. Later, you will review the basic concepts behind DevOps and how to extend the GitLab DevOps platform with Comet, further enhancing your ability to deploy your data science projects. Finally, you will cover various use cases of Comet in machine learning, NLP, deep learning, and time series analysis, gaining hands-on experience with some of the most interesting and valuable data science techniques available. By the end of this book, you will be able to confidently build data science pipelines according to bespoke specifications and manage them through Comet.
Table of Contents (16 chapters)
1
Section 1 – Getting Started with Comet
5
Section 2 – A Deep Dive into Comet
10
Section 3 – Examples and Use Cases

First use case – offline and existing experiments

In Chapter 1, An Overview of Comet, you built a simple use case that permitted you to track images in Comet. The example used 52 time series indicators related to gross domestic product (GDP) in Italy, built 52 images, and uploaded them to Comet.

During the experiment, you will surely have noticed that the loading of the images in Comet was quite slow, depending on the bandwidth available in your internet connection. In this example, you will see how to use the concepts of offline and existing experiments to make the loading process smoother. You will also see how an existing experiment can be improved at a later time.

In this example, we suppose that the code implemented in Chapter 1, An Overview of Comet, for the first use case is running. Thus, please refer to it for further details.

The full code of this example is available in the GitHub repository, at the following link: https://github.com/PacktPublishing/Comet...