Categories

Versions

You are viewing the RapidMiner Studio documentation for version 9.5 - Check here for latest version

What's New in RapidMiner Studio 9.4.1?

Released: Sep 25th, 2019

The following describes the bug fixes in RapidMiner Studio 9.4.1:

New Features

  • Model Ops: new view in Studio which allows to deploy and manage models
    • One-click deployment makes model deployment do-able for anyone of any skill level
    • Organize models in deployments across multiple locations
    • Share models with other users
    • Simple and robust scoring options for your models
    • Model leaderboards display performance of champion and challenger models in real-time
    • Advanced visualizations help you understand performance and business impact
    • Monitoring and alerts prevent model drift and bias
    • Integrations and governance features help you with model web services and explain how the models are being used
  • Auto Model: highlight models with largest profits and business impact
  • Added new connection type Mail (send) that replaces the mail configuration in the settings
  • Used new mail connections in Send Mail and Root operator

Enhancements

  • Improved runtime of the Random Forest operator when using least squares in a regression scenario.
  • Added parameter to Stacking operator that allows to train the stacked model on the confidence values of the base learners.
  • Result tabs no longer switch positions if more than one row of results is displayed.
  • Time Series: Added log values to the Sliding Window Validation operator
  • Time Series: Added UserErros for STL Decomposition operator in case the input data has missing or infinite values

Bugfixes

  • Fixed yet another problem that could lead to RapidMiner Studio starting with a size of 0x0 pixels on OS X
  • Visualizations: Fixed y axis max value setting sometimes being ignored when more than one y axis was present
  • Improved errors in case of many attributes for the Correlation Matrix operator.
  • Time Series: Fixed (made it consistent) the order of models in the Multi Label Model and the order of performances in the collection output of the Multi Label Performance operator
  • Time Series: Fixed meta data of Process Windows output ports (now output is correctly recognized as a collection)

Development

  • Faulty listeners for Studio preference changes no longer prevent subsequent listeners from getting informed in case of an exception in the previous listener