Apache Mahout Samsara: The Quick Start

Last week the newest Apache Mahout 0.10 was released. One of the new features it has is a new math environment called “Samsara”, or Mahout Scala/Spark Bindings.

Samsara is a Linear Algebra library for Mahout. It’s written in Scala, which makes it possible to use operator overloading and it features nice R-like or Matlab-like syntax for basic Linear Algebra operations. For example, matrix multiplication is just X %*% Y. What is more, these operations can be distributed and run by an executing environment - currently by Apache Spark.

In this article we will see how to quickly set up a basic skeleton project and then we’ll try to do some very simple analysis on a 200 MB dataset.

Naive Bayes on Apache Flink

In this blog post we are going to implement a Naive Bayes classifier in Apache Flink. We are going to use it for text classification by applying it to the 20 Newsgroup dataset. To understand what is going on, you should be familiar with Java and know what MapReduce is. If you have seen and understood a word count example in any system, you're good to go. If you haven't heard of MapReduce or haven't seen the word count, you may first have a look at our introductory post "Hadoop and MapReduce".

Hadoop and MapReduce

In this article we will briefly discuss the computation paradigm MapReduce, and Apache Hadoop as one of its implementations. We won't get into much details, and we even won't implement the Word Count on Hadoop, but it should give some foundation for the future articles about tools for scalable data processing.