You are viewing the RapidMiner Studio documentation for version 9.4 - Check here for latest version
What's new in RapidMiner Studio 9.4.1
New Automated Model Ops
Follow the fully automated data science path: prepare your data using Turbo Prep, create prediction models via Auto Model and finally put them into production with Model Ops.
- Deploy the most promising models with one click and score new data via flexible web services or in the UI.
- Track model performance on an intuitive dashboard and swap easily to the best performing one. Setup an email alert to get notified if a model outperforms the one in production.
- Evaluate each model with respect to their financial impact instead of pure Data Science metrics.
- Detect changes in data and their impact on model performance early to address problems.
- Use our integrated dashboard to keep track of data drift and model performance.
New map visualizations
Visualize geospatial data with the new map visualizations. You can choose from multiple map types with many different configuration options, as well as dozens of maps for geographic regions, continents, and countries. Available map types:
- Choropleth maps: Used to display numeric values associated to regions (e.g. a country or a state) via a color gradient
- Categorical maps: Used to visualize regions that belong to a number of distinct categories
- Point maps: These maps offer latitude and longitude support and display a marker for each coordinate on the selected map
Three new chart types have been added in addition to some tweaks and fixes to the existing charts:
- Sunburst chart
- Chord diagram
- Parliament chart
Improved Auto Model
Auto Model features several improvements under the hood as well as a few more visible enhancements:
- All predictive processes generated by Auto Model are now much cleaner, well-structured, and can be understood way easier.
- Cost-sensitive learning has been added to show the costs / benefits in the validation result. This allows to solve problems (e.g. fraud detection) that involve highly imbalanced data sets (e.g. credit card transaction data).
New data prep and modeling capabilities
Several new operators have been added to ease and enhance data preparation and machine learning:
- New operators Replace All Missings, Handle Unknown Values, One Hot Encoding and Append (Robust) to easily prepare data for modeling and scoring.
- New operator Rescale Confidences (Logistic) to rescale confidences even for classification with more than two classes.
- New operator Cost-Sensitive Scoring: Novel approach for cost-sensitive learning which works for more than two classes.
- New operators Multi Label Modeling and Multi Label Performance to train and validate a combined model for multiple label columns in a single step.
Enhanced time series forecasting
New operators have been added for
- Forecasting multiple horizons of a time series with any machine learning model (Multi Horizon Forecast)
- Validating performance of multi horizon forecasts (Multi Horizon Performance)
- Sliding window validation for time series data science problems
Enhanced data source connection framework
All RapidMiner-supported connectivity extensions on the Marketplace now use the new data source connection framework, which includes handling connections to
Enhancements and bug fixes
The following pages describe the enhancements and bug fixes in RapidMiner Studio 9.4.1 releases: