You are viewing the RapidMiner Radoop documentation for version 7.6 - Check here for latest version

What’s New in RapidMiner Radoop 7.0?

Enhancements and bug fixes

The following improvements are part of RapidMiner Radoop 7.0.

Added Single Process Pushdown operator
Added Decision Tree distributed learner operator (integrates Spark ML)
Added Random Forest distributed learner operator (integrates Spark ML)
Added Support Vector Machine distributed learner operator (integrates Spark MLlib)
Generate Attributes now has a new intuitive UI, that can also be used in Filter Examples operator and on the Hadoop Data view as well
Added several new functions of Hive to the Generate Attributes expression editor
Adds support for Spark 1.6.0
Applied new operator grouping and new color for Radoop operators
Added an XML editor for Radoop connections on the Advanced Connection Properties dialog
Added new convergence tolerance parameter to Linear Regression and Logistic Regression
Added new connection test: upload jar file required for user-defined functions to the cluster
Added new connection test: create permanent user-defined functions
Added new connection test: test Spark staging directory
Custom JDBC driver for Hive can now be specified beside the built-in drivers
Importing large tables into the cluster is now faster
Using Tez as the execution engine for Hive (Hive-on-Tez) is now possible, although not yet supported

BUGFIX: Local temporary files are now always cleaned
BUGFIX: Pivot now does not truncate the average value of integer attributes
BUGFIX: Temporary tables and directories are now also cleaned if a special error occurs
BUGFIX: Hive command timeout now applied to each command instead of multiple commands in some cases
BUGFIX: In Replace Missing operator real values are now not allowed for integer attributes
BUGFIX: Job kill for certain operators now works on secure cluster as well
BUGFIX: Hadoop Data view related panels are now not listed on the Show Panel menu
BUGFIX: Fixes issues with non-default Hive null placeholder setting
BUGFIX: Long error messages now do not appear in pop-up windows that are too big
BUGFIX: Move into new subprocess action on the Design canvas now adds Radoop Subprocess operator
BUGFIX: Distributed model scoring on a new cluster or with new user no longer fails when running from RM Server