You are viewing the RapidMiner Studio documentation for version 8.0 - Check here for latest version
What's new in RapidMiner Studio 8.0
Update / migration
If you use RapidMiner Server, please make sure the latest version RapidMiner Server 8.0 is installed before you upgrade your Studio, as it is not backwards compatible. Update is available through the RapidMiner Marketplace.
Parallelized model optimization operators
The operators Optimize Parameters (Grid) and Loop Parameters now run in parallel leveraging multiple CPU cores (depending on license type). This translates directly into faster processes and improved performance when tuning hyperparameters of machine learning models.
Please be aware that the old operators are deprecated. Existing processes containing them will still use the previous versions and they need to be replaced by hand in order to benefit from the parallelization.
Regression Trees and further Decision Tree enhancements
We enhanced the capabilities of our Decision Tree and Random Forest algorithms:
- Regression Trees - Our Decision Tree and Random Forest operators can now handle numerical labels and solve regression problems as well. A new splitting criterion based on the least-squared statistic has been added for these tasks.
- Extremely Randomized Trees - For Random Forest, there is a new option to select splits randomly which helps to build more robust trees and to prevent overfitting. When selected, cuts for numerical attributes are done randomly.
- Provision of feature weights - The Decision Tree and Random Forest operators now provide a new port that outputs feature weights, capturing and allowing to quantify the importance of attributes when building the tree.
Fuzzy operator search
Searching has never been easier! Now our operator search is fuzzy, which means that you can misspell a bit and still find what you are looking for.
Improved operator help
We have continued to improve the help texts of the RapidMiner operators. A new set of operators has been reviewed for and improved for this release to make it easier for new users to start working with RapidMiner Studio and to help experienced users get acquainted with any operator more easily.
This is the list of improved operators:
- Naïve Bayes
- Normalize
- k-Means
- k-NN
- Join
- Performance (Binominal Classification)
Enhancements and bug fixes
The following pages describe the enhancements and bug fixes in RapidMiner Studio 8.0 releases: