Preface

A recent survey of machine learning professionals (https://www.comet.com/site/about-us/news-and-events/press-releases/comet-releases-new-survey-highlighting-ais-latest-challenges-too-much-friction-too-little-ml/) concluded that about 40%–60% of interviewed professionals abandoned their data science projects because they were not able to manage the full life cycle process of their data science projects. I’m a data science researcher, and before encountering Comet, I belonged to that 40%–60% of professionals who abandon their data science projects. In fact, during my working experience, I have abandoned many projects without concluding them because of the nature of research, where you test an idea and, if it does not work, you drop it.

Almost a year ago, I discovered Comet, a platform for model tracking and monitoring, and some wonderful people from its team, who opened my mind to the many features provided by Comet. I began to study it, with the hope of keeping my projects organized and moving them from early stages to production. I realized that I was able to conclude all the projects I implemented in Comet because of the simplicity of the platform.

Comet for Data Science is the result of my studies and tests, as well as the countless biweekly meetings with the Comet team. The book aims at helping you to learn how to manage a data science project workflow, from its early stages up to project deployment and reporting. In a single sentence, Comet for Data Science is written to help you to conclude your data science projects successfully.

By picking this book, you will look at the general concepts of data science from a Comet perspective, with the hope that you will increase your productivity. The book will take you through the journey of building a data science project and integrating it into Comet, including exploratory data analysis, model building and evaluation, report building, and, finally, moving the model to production. Throughout the book, you will implement many practical examples that you can use to better understand the described concepts, as well as starting points for your projects.

I hope that this book will add something to your knowledge, and – why not? – help you to become a better data scientist!

Happy reading!