You are viewing the RapidMiner Studio documentation for version 7.6 - Check here for latest version
What's New in RapidMiner Studio 7.3.0?
The following describes the enhancements and bug fixes in RapidMiner Studio 7.3.0:
Enhancements
- New parallel Cross Validation operator replaces X-Validation, Batch X-Validation, and X-Prediction.
- Operator search now also searches for matching Marketplace extensions
- Greatly improved Proxy UI and logic
- Logistic Regression, Generalized Linear Model and Gradient Boosted Trees now return Attribute Weights output as well
- Added reproducible parameter to Logistic Regression, Generalized Linear Model and Gradient Boosted Trees. If checked, the result is guaranteed to be the same, because the parallelization level is fixed.
- Improved sorting for repository entries.
- Performance improvement for Rule Induction and Perceptron operators.
- Improved high DPI support.
- Improved operator progress for Apply Model and Logistic Regression (SVM).
- Improved welcome dialog layout.
Bug fixes
- Fixed NullPointerException in Logistic Regression and Generalized Linear Model with compute p-values on and solver set to AUTO on an input with large number of nominal values
- Changed the default of the max_w2 parameter of Deep Learning to 10, as the operator help describes; it also became a non-advanced parameter
- Fixed some minor tutorial inconsistencies
- If there is a security error, Logistic Regression, Generalized Linear Model, Gradient Boosted Trees and Deep Learning operators can recover without Studio / Server restart
- Input data rebalancing in Logistic Regression, Generalized Linear Model, Gradient Boosted Trees and Deep Learning no longer depends on the number of cores but the number of threads (configurable)
- Logistic Regression, Generalized Linear Model, Gradient Boosted Trees and Deep Learning operators are now loaded even if javafx package is missing from the Java Runtime Environment
- Fixed multiple problems with the GSP operator
- Operator progress now vanishes if operator is successfully stopped
- Fixed operator progress animation being stuck sometimes
- Fixed import excel data UI issues on Mac OS X
- Fixed that in-Hadoop scoring of Logistic Regression, Generalized Linear Model, Gradient Boosted Trees and Deep Learning models in Rapidminer Radoop no longer logs something for each row (leads to significant performance improvement)
Development
- Added a centralized API for data table creation: From now on a new ExampleSet should be created via an ExampleSetBuilder provided by the ExampleSets class instead of using MemoryExampleTable
- Tweaked project structure for the open source core. This does not affect the functionality of RapidMiner Studio.
