Apache Mahout : Beyond Mapreduce

Apache Mahout: Beyond MapReduce. Distributed algorithm design

This book is about designing mathematical and Machine Learning algorithms using the Apache Mahout "Samsara" platform.

The material takes on best programming practices as well as conceptual approaches to attacking Machine Learning problems in big datasets. Math is explained, followed by code examples of distributed and in-memory computations.

Written by Apache Mahout committers for people who want to learn how to design distributed math algorithms as well as how to use some of the new Mahout "Samsara" algorithms off-the-shelf.

The book covers Apache Mahout 0.10 and 0.11.

"A certain level of mathematical understanding is required to get a feeling for how and why these algorithms work and how they can benefit the user. However, too deep a treatment loses sight of the higher level goals of the problems we are actually trying to solve. This book strikes that balance enabling users to assemble more complex algorithms and solve problems."
            - Nathan Halko
"The definitive guide to the new Mahout 'Samsara' distributed linear algebra framework. This book contains just the right amount of math needed to implement complex algorithms without having to worry about scalability."
             - Suneel Marthi
"This book is a welcome introduction or refresher for people interested in the math-oriented work that goes into doing linear algebra at scale. Often texts and tutorials in this area focus on building up motivation from the math side or on how to use existing machine-learning libraries and tools as black boxes. Taking a different approach, the authors of this book lead the reader down the path of using Apache Mahout as a workbench for experimenting with vector math on very large data sets, introducing a syntactically lightweight and straightforward approach that will be familiar to users of R and other math-focused languages."
              - Andrew Musselman