What this book covers
Chapter 1, An Overview of Comet, is a general introduction to Comet, an experimentation platform, which allows you to manage and optimize machine learning projects, from their early stages to their final deployment. First, you will learn what Comet is and who its target users are. Then, you will get familiar with the Comet basic concepts, including projects, experiments, workspaces, and panels. Finally, you will build two basic use cases in Comet.
Chapter 2, Exploratory Data Analysis in Comet, guides you to use Comet to perform Exploratory Data Analysis (EDA). First, you will be introduced to the main steps to perform EDA, including problem setting, data preparation, preliminary data analysis, and preliminary results. Then, you will review the main two techniques used to perform EDA: visual and non-visual EDA. Finally, you will learn how to use Comet for EDA through a practical example.
Chapter 3, Model Evaluation in Comet, guides you to use Comet to perform model evaluation. First, you will be introduced to the main concepts to evaluate the performance of a model, such as data splitting, how to choose metrics for evaluation, and the basic concepts behind error analysis. Then, you will see the main model evaluation techniques for different data science tasks (classification, regression, and clustering). Finally, you will learn how to use Comet for model evaluation, through a practical example.
Chapter 4, Workspaces, Projects, Experiments, and Models, deepens some concepts regarding Comet. First, you will see some advanced concepts on workspaces, projects, and experiments, as well as how to perform parameter optimization in Comet. Then, you will learn how to implement a Comet experiment using R or Java as the main programming language. Finally, you will extend the basic examples implemented in Chapter 1, An Overview of Comet.
Chapter 5, Building a Narrative in Comet, describes some strategies to build a good report in Comet. First, you will review the basic concepts and techniques to build a narrative from data, including an overview of the DIKW pyramid, and how to turn your data into a story. Then, you will learn how to build a narrative in Comet through two practical examples.
Chapter 6, Integrating Comet into DevOps, provides you with practical concepts and examples on DevOps and MLOps and how to integrate them into Comet. First, you will review the basic concepts and best practices related to DevOps and MLOps. Then, you will learn how to integrate Comet into the DevOps/MLOps paradigm through the concept of the REST API. Next, you will analyze Docker and Kubernetes, two of the most common frameworks for DevOps. Finally, you will learn how to integrate Comet in Docker and Kubernetes through two practical examples.
Chapter 7, Extending the GitLab DevOps Platform with Comet, describes the concept of Continuous Integration (CI) and Continuous Delivery (CD), how to implement it using GitLab, and how to integrate Comet in a CI/CD workflow. First, you will review the basic concepts of CI/CD, including the CI/CD workflow and the concept of a source control system. Then, you will see the GitLab basic concepts, including an overview of its architecture, how versioning works, and the basic GitLab commands.
Then, you will configure GitLab to work with Comet. Finally, you will see a practical example that will help you to get familiar with the described concepts.
Chapter 8, Comet for Machine Learning, provides you with an overview of the Machine Learning (ML) concepts, with a focus on the scikit-learn library, and how to integrate them in Comet. First, you will review the basic ML concepts, including a classification of the main ML systems, the main ML models, and their main challenges. Then, you will review the scikit-learn package, with a focus on preprocessing, dimensionality reduction, model selection, supervised learning, and unsupervised learning. Finally, you will learn how to integrate Comet with scikit-learn through a practical example.
Chapter 9, Comet for Natural Language Processing, illustrates the main concepts behind Natural Language Processing (NLP), with a focus on the Spark NLP library, and how to integrate the main concepts in Comet. First, you will review the basic NLP workflow and also learn how to classify the main NLP systems and what their main challenges are. Then, you will review the Spark NLP library, including the concepts of annotation and pipeline. Finally, you will learn how to integrate Comet with Spark NLP through a practical example.
Chapter 10, Comet for Deep Learning, describes the main concepts behind Deep Learning (DL), with a focus on the TensorFlow library, and how to integrate them in Comet. First, you will review the basic concepts behind neural networks, their difference with respect to DL networks, and how to classify DL networks. Then, you will review the TensorFlow library, with a focus on how to load a dataset, as well as how to build a train a model. Finally, you will learn how to integrate Comet with TensorFlow through a practical example.
Chapter 11, Comet for Time Series Analysis, reviews the main concepts related to Time Series Analysis (TSA), with a focus on the Prophet library, and how to integrate them into Comet. First, you will review the basic concepts behind TSA, including the concept of stationarity, the time series components, and how to check the presence of breakpoints in a time series. Then, you will be introduced to the Prophet library, with a focus on how to build a prediction model. Finally, you will learn how to integrate Comet with Prophet through a practical example.