Chapter 1. Introduction to Mahout

Mahout is an open source machine learning library from Apache. Mahout primarily implements clustering, recommender engines (collaborative filtering), classification, and dimensionality reduction algorithms but is not limited to these.

The aim of Mahout is to provide a scalable implementation of commonly used machine learning algorithms. Mahout is the machine learning tool of choice if the data to be used is large. What we generally mean by large is that the data cannot be processed on a single machine. With Big Data becoming an important focus area, Mahout fulfils the need for a machine learning tool that can scale beyond a single machine. The focus on scalability differentiates Mahout from other tools such as R, Weka, and so on.

The learning implementations in Mahout are written in Java, and major portions, but not all, are built upon Apache's Hadoop distributed computation project using the MapReduce paradigm. Efforts are on to build Mahout on Apache Spark using Scala DSL. Programs written in Scala DSL will be automatically optimized and executed in parallel on Apache Spark. Commits of new algorithms in MapReduce have been stopped and the existing MapReduce implementation will be supported.

The purpose of this chapter is to understand the fundamental concepts behind Mahout. In particular, we will cover the following topics: