On Aug 10, 2018, Apache MADlib completed its fourth release as an Apache Software Foundation Top Level Project.
New features include: * Utilities - Columns to vector, vector to columns, drop columns. Improvements: * Multilayer perceptron - Added momentum and Nesterov's accelerated gradient methods to gradient updates. * Statistics - Added grouping support to correlation and covariance. * Decision tree/random forest - Added impurity variable importance. * Decision tree/random forest - Added new helper function to report variable importance values in a more readable way. * Install - Refactored and updated the madpack installation and upgrade tool. You are invited to download the 1.15 release https://archive.apache.org/dist/madlib/1.15/ and review the release notes https://github.com/apache/madlib/blob/master/RELEASE_NOTES Thanks to all of those in the diverse Apache MADlib community who contributed to this release! Aaron VP of Apache MADlib