Parallel Python with Dask: Perform distributed computing, concurrent programming and manage large dataset

by Tim Peters

Dask has revolutionized parallel computing for Python, empowering data scientists to accelerate their workflows. This comprehensive guide unravels the intricacies of Dask to help you harness its capabilities for machine learning and data analysis. Across 10 chapters, you'll master Dask's fundamentals, architecture, and integration with Python's scientific computing ecosystem. Step-by-step tutorials demonstrate parallel mapping, task scheduling, and leveraging Dask arrays for NumPy workloads. You'll discover how Dask seamlessly scales Pandas, Scikit-Learn, PyTorch, and other libraries for large datasets. Dedicated chapters explore scaling regression, classification, hyperparameter tuning, feature engineering, and more with clear examples. You'll also learn to tap into the power of GPUs with Dask, RAPIDS, and Google JAX for orders of magnitude speedups. This book places special emphasis on practical use cases related to scalability and distributed computing. You'll learn Dask patterns for cluster computing, managing resources efficiently, and robust data pipelines. The advanced chapters on DaskML and deep learning showcase how to build scalable models with PyTorch and TensorFlow. With this book, you'll gain practical skills Packed with hands-on examples and expert insights, this book provides the complete toolkit to harness Dask's capabilities. It will empower Python programmers, data scientists, and machine learning engineers to achieve faster workflows and operationalize parallel computing.

172 pages, Paperback

Published October 19, 2023

Reviews

[John · Rating 3 out of 5]

This is cheap but clear and comprehensive coverage of Dask. It's a bit repetitive, there are obviously cases where the wrong examples were cut and pasted too many times, and occasionally the copy editing is not great. However, it does show Dask in about as many different cross-library situations as you're probably going to want to use it with good, clear framing and (most of the time) salient code examples.

Are the more expensive books going to be sufficiently better to justify the cost? It's unlikely I'm going to bother to find out.