Xiangrui Meng, Doris Xin, Paul Ogilvie and Jonathan Traupman

In the BigData Innovators Gathering of World Wide Web Conference 2014 (Big 2014)


 

Abstract

Researchers applying machine learning to large data sets commonly focus on scalable optimization algorithms. However, applying machine learning to improve data products requires a wide array of scalable algorithms that have been under-investigated. Based on our experience working with data scientists and research engineers at LinkedIn, we describe the wider set of considerations required to build complete data products. We also describe Metronome, a framework for processing big data we have developed at LinkedIn.