Book Image

Mastering Apache Spark 2.x - Second Edition

Book Image

Mastering Apache Spark 2.x - Second Edition

Overview of this book

Apache Spark is an in-memory, cluster-based Big Data processing system that provides a wide range of functionalities such as graph processing, machine learning, stream processing, and more. This book will take your knowledge of Apache Spark to the next level by teaching you how to expand Spark’s functionality and build your data flows and machine/deep learning programs on top of the platform. The book starts with a quick overview of the Apache Spark ecosystem, and introduces you to the new features and capabilities in Apache Spark 2.x. You will then work with the different modules in Apache Spark such as interactive querying with Spark SQL, using DataFrames and DataSets effectively, streaming analytics with Spark Streaming, and performing machine learning and deep learning on Spark using MLlib and external tools such as H20 and Deeplearning4j. The book also contains chapters on efficient graph processing, memory management and using Apache Spark on the cloud. By the end of this book, you will have all the necessary information to master Apache Spark, and use it efficiently for Big Data processing and analytics.
Table of Contents (21 chapters)
Title Page
Credits
About the Author
About the Reviewer
www.PacktPub.com
Customer Feedback
Preface
10
Deep Learning on Apache Spark with DeepLearning4j and H2O

Classification with Naive Bayes


This section will provide a working example of the Apache Spark MLlib Naive Bayes algorithm. It will describe the theory behind the algorithm and will provide a step-by-step example in Scala to show how the algorithm may be used.

Theory on Classification

In order to use the Naive Bayes algorithm to classify a dataset, the data must be linearly divisible; that is, the classes within the data must be linearly divisible by class boundaries. The following figure visually explains this with three datasets and two class boundaries shown via the dotted lines:

Naive Bayes assumes that the features (or dimensions) within a dataset are independent of one another; that is, they have no effect on each other. The following example considers the classification of e-mails as spam. If you have 100 e-mails, then perform the following:

60% of emails are spam
80% of spam emails contain the word buy
20% of spam emails don't contain the word buy
40% of emails are not spam
10% of non...