MADlib - Big Data Machine Learning the Apache Way

MADlib is a machine learning library for data scientists. The MAD stands for "Magnetic, Agile and Deep". The concept is doing big data analytics in the database. Another key principle of MADlib is leverage of MPP share nothing architectures, first elucidated on by Michael Stonebraker at University of California, Berkeley. Joe Hellerstein at UCB is a big promoter of MADlib.