Part 3. Classification

This third and final part of Mahout in Action, comprising chapters 13 through 17, covers how to use Mahout for classification. Using the techniques presented here, you’ll be able to structure questions and choose and prepare data appropriately to have machines automatically assign data to preselected categories. Classification is a simplified form of decision making that gives discrete answers to an individual question. Machine-based classification is an automation of this decision making process that learns from examples of correct decision making and emulates those decisions automatically—a core concept in predictive analytics. Classification’s reliance on guided learning and focus on answering one question at a time distinguish it from clustering and recommendation, discussed in the previous two parts of this book. Clustering, in contrast to classification, relies on machines to decide on their own; recommenders select and rank the best of many possible answers.

This part of the book takes you through three stages in classification. Chapters 13 and 14 introduce the basics and show you how to build and train a Mahout classification model. Chapter 15 explains how to use evaluation throughout the process to identify the best versions of your model and fine-tune it for best performance. Chapters 16 and 17 show you how to deploy your classification system in real-world situations. Techniques to correctly identify and extract useful data are presented, along with detailed discussions of how to optimize the features as vectors usable by the various Mahout classification algorithms. Tips for avoiding problems such as target leaks are also included. Step-by-step examples give you experience with each step to help you build classifiers and fine-tune them for best performance.

In addition to the detailed explanation of how to build and deploy an effective and reliable classifier for systems with huge requirements for speed and scale, we present in chapter 17 a real-world case study of an online marketing company that uses Mahout classification to address its needs. This final real-world case study lets you see how the ideas presented throughout the classification chapters can be applied.