You are viewing the RapidMiner Radoop documentation for version 7.6 - Check here for latest version
What’s New in RapidMiner Radoop 2.6?
This page describes the new features as well as enhancements and bug fixes available in RapidMiner Radoop 2.6.
Spark scripting in Python and R
RapidMiner Radoop 2.6 now includes a Spark Script operator that can execute arbitrary PySpark and SparkR scripts as part of the workflow. Developers can make use of the convenient SparkSQL and Dataframe API to write highly effective Spark programs in case they do not find any RapidMiner Radoop code-free operators for their particular task.
Currently, the Spark Script operator only works with Spark 1.5 and does not support kerberized Hadoop clusters.
Marketplace delivery and easier licensing
Starting from RapidMiner Radoop 2.6, it is delivered through the RapidMiner Marketplace which makes installation and updates much easier. Also, Radoop has been integrated into the licensing framework of the whole RapidMiner platform so licenses are now delivered the same way as Studio and Server licenses, and the licenses become readily available and applied automatically after logging in inside RapidMiner Studio. Old Radoop license files no longer work with RapidMiner Radoop 2.6.
New operators for data preprocessing
This release adds several convenience operators for bulk renaming (Rename by Replacing, Rename by Generic Names) and converting attributes to numerical so they can be used in learning algorithms (clustering or classification) that only accept numerical attributes. This new Nominal to Numerical operator supports dummy coding and conversion to unique integers.
Enhancements and bug fixes
The following improvements are part of RapidMiner Radoop 2.6.
Enhancements
- Improved error messages in case of connection issues
- Improved process validation for optional parameters
- Improved error messages for Pig Script operator when the result has an unknown schema
- Improvements to Declare Missing Values operator
Bug fixes
- BUGFIX: Generate Attributes now handles binominals properly in case of overwrite mode
- BUGFIX: Connection Dialog no longer asks to overwrite the connection when nothing has been changed