You are viewing the RapidMiner Radoop documentation for version 7.6 - Check here for latest version
What’s New in RapidMiner Radoop 7.0?
Versions and compatibility
Please note that starting from version 7.0 the complete RapidMiner platform now has a unified version numbering, i.e. products like RapidMiner Server or RapidMiner Radoop will use the same version 7.0 from now on.
New distributed Machine Learning algorithms
RapidMiner Radoop 7.0 adds three new machine learning algorithms based on Spark, extending the number of distributed modeling operators to 12.
- New Decision Tree operator for Spark 1.5+ that supports multinominal classification as well. The old binary Decision Tree remains supported until most clusters upgrade to Spark 1.5+.
- New Random Forest operator for Spark 1.5+ with improved confidence voting, hence delivers better results than pure Spark implementations.
- New Support Vector Machine operator for all supported Spark versions.
Single Process Pushdown
Starting from RapidMiner Radoop 7.0, the Single Process Pushdown operator is available to push down processing of practically all RapidMiner functions so they can be computed inside one of the Hadoop nodes. It is packaging up the RapidMiner engine, necessary extensions, and the process into a Spark job and pushing it into one of the Hadoop nodes for execution. Now users can combine the powerful native Hadoop operations with the vast amounts of data prep and modeling functions available in RapidMiner - all powered by the same Hadoop cluster. This feature complies with Hadoop resource management and security policies and can be used without any manual installation steps.
New look & feel
RapidMiner Radoop 7.0 also comes with the completely modernized look & feel greatly improving the usability of the platform. Important functions stay front and center while less frequently used features have been moved out of the way. It also includes the new expression builder introduced in RapidMiner Studio 6.5 and it adds extended help for all available Hive functions.
In summary, these changes provide a cleaner design to make it easy for everyone to generate predictive workflows without sacrificing deep functionality required by experienced users.
Enhancements and bug fixes
The following pages describe the enhancements and bug fixes in RapidMiner Radoop 7.0 releases: