Home

Metronome is a suite of parallel iterative algorithms that run natively on Hadoop's Next Generation YARN platform.

• Based directly on work we did with Knitting Boar and [IterativeReduce] (https://github.com/emsixteeen/IterativeReduce)
  • Knitting Boar provided a base implementation of Parallel logistic regression on parallel SGD, showing linear scale with input size
  • Built on top of BSP-style computation framework "Iterative Reduce" (Hadoop / YARN)
  • Uses Mahout's implementation of Stochastic Gradient Descent (SGD) as basis for worker process
• Linear Regression package can produce a linear regression model off large amounts of data
• Packaged in a new suite of parallel iterative algorithms called Metronome
  • 100% Java, ASF 2.0 Licensed, on github

Original Presentation from Hadoop Summit EU 2013

<iframe src="http://www.slideshare.net/slideshow/embed_code/17636499" width="427" height="356" frameborder="0" marginwidth="0" marginheight="0" scrolling="no" style="border:1px solid #CCC;border-width:1px 1px 0;margin-bottom:5px" allowfullscreen webkitallowfullscreen mozallowfullscreen> </iframe>

Hadoop Summit EU 2013: Parallel Linear Regression, IterativeReduce, and YARN from Josh Patterson

Resources

• [IterativeReduce Programming Model] (https://github.com/emsixteeen/IterativeReduce/wiki/Iterative-Reduce-Programming-Guide)
• Using IRUnit - the IterativeReduce Unit Testing Framework
• Running parallel linear regression with IRUnit on synthetic test data

Contributors and Special Thanks

• Michael Katzenellenbollen
• Dr. Jason Baldridge
• Dr. James Scott