Altair HyperStudy 2020 Release Notes

New Features

Process flow visualization: The new Block View tool enables you to save the current view of your model for future reference and recall saved views.
Customizable Fits: Users can now link HyperStudy to an external predictive model of their choice to use in the Fit approach. This is called external fit registered by the user. Both regression and classification types of algorithms are supported.
Predictive modeling support for classification: The new fit method, Random Forest, has been introduced for classification problems. Random Forest is a classifier algorithm formed of many decision trees. Random Forest is quite fast and can deal with unbalanced and missing data.
Deploy predictive models created in Altair Knowledge Studio: There is a new model type called Knowledge Studio. Knowledge Studio’s predictive models such as Deep Learning, Logistic Regression, and so on, can easily be deployed within HyperStudy for optimization problems.
Automatic detection of variable names within the Editor: The Editor Tool now detects variable names in parameterized text files. When a value is selected to be parameterized, the Editor Tool in HyperStudy automatically assigns the Label the value name.

New visualization of Fit accuracy: In the Residual Tab, users can now plot the residuals of Input, Cross-Validation, and Testing Matrices in addition to table view.
Simplified process to share data between approaches: In the Specification step of each approach, there is now a dedicated window to help users conveniently share data between approaches or import external data.
Modify the target of existing model resources: To easily replace a file in Model Resources, the Modify Resource option has been added to the right-click context menu.
Easily register custom reports: Register custom report generators through *.mvw preference files.
Direct reader for hstp format: The new data source tool, Hstp Reader, enables you to directly extract results from software that provides hstp in ASCII format.
Scale factor in curve matching problems: The AREA tool in Data Source now has scale factors available in the GUI.

The default value for the maximum evaluations in a multi-objective GRSM optimization has been reduced from 500 to 200.

Future versions of HyperStudy will remove support for upgrading studies earlier than version 12.0.