Model Menu (ASCMO-STATIC)

The Model menu consists of the following entries:

Start ASC GP Model Training

Here you start the training of an ASC GP model (see Model Types).

Model Properties

Opens the Model Properties (ASCMO-STATIC) window where you can adjust the settings and the transformation of model inputs to be used for the respective model (output).

Automated Machine Learning

Opens the Automated Machine Learning window, which provides the possibility to use machine learning models and techniques without experts knowledge in machine learning.

Valid Model Range

Note
* This item is only available when the Advanced Settings are enabled.

Opens the "Valid Model Range" window where you can set a threshold (absolute value of the maximum standard deviation) with which the valid range of the model (output) is determined. The range is displayed with the Show Valid Model Range option in the View menu). Default: Standard deviation of the training data point with the biggest deviation.

Reference Model *

Set as Reference > *	Copies the current model to reference model for a selected output (* = <output_name>) or all outputs (* = All Outputs).
Reset to Reference > *	Resets the current model for a selected output or all outputs to the previously defined reference model.
Delete Reference > *	Deletes the reference model for a selected output (* = <output_name>) or all outputs (* = All Outputs).
Show Reference Model> *	Shows/hides the reference model information in the main interface. The reference model for a selected output or all outputs can only be displayed if a model has been set as reference previously.

Error (Leave-One-Out)

For the analysis based on leave-one-out error, a model is formed on all training data except for one measurement. In this case, this omitted measuring point represents the test data with which the model prediction is tested.

Measured vs. Predicted	Opens the "Measured vs. Predicted (Leave-One-Out)" window.
Show in ISP *	Displays the absolute errors of the measuring points (referenced to the current model prediction) in the ISP. This display allows a qualitative assessment of how trustworthy the model prediction is in this range (of the input variable or the current operating point).
Errors vs. Output	Opens the "Errors vs. Output (Leave-One-Out)" window.
Probability Plot	Opens the "Probability Plot (Leave-One-Out)" window.
Error vs. Input > *	Opens the "Error vs. Input" window. The sub menu contains each input and, in addition, Open All.
Error vs. Run Order	Opens the "Error vs. Run Order" window.
Error over Training Data Size	Opens the "Analyze model Leave-One-Out error for <output>" window.

Error (Test Data)

For the analysis based on test data, the model outputs are compared with the test data. Test data are data which the model did not use for training. This makes it possible, e.g. to assess the capability of the model for generalization.

Otherwise, the functions of the menu correspond to those of the Error (Leave-One-Out) menu, except that Error over Training Data Size is not available.

Error (Training Data)

The third option is the analysis based on training data. In the process, the deviation between the measured values at all training data from the model output is compared at the corresponding location and graphically displayed.

Otherwise, the functions of the menu correspond to those of the Error (Leave-One-Out) menu, except that Error over Training Data Size is not available.

Model Compression

Note
The "Model Compression" function (Model > Model Compression) is available as an add-on to ETAS ASCMO, for which the special license "ASCMO_COMPRESSION_INTERNAL" is required. The Model Compression menu item is only visible if you have enabled the Advanced Settings.

Note

The "Model Compression" function (Model > Model Compression) is available as an add-on to ETAS ASCMO, for which the special license "ASCMO_COMPRESSION_INTERNAL" is required.

The Model Compression menu item is only visible if you have enabled the Advanced Settings.

The "Model Compression" function (Model > Model Compression) allows you to keep the speed of ASCMO-STATIC on the usual level even for huge datasets - especially for the integrated optimizers. It reduces in a smart way the number of used data points (model size) and with it the complexity of the models. With this Add-On, you can use the compressed models within ASCMO-STATIC.

Before you can use the functions of compressed models, you must first define the model type "Compressed Model" for the desired output in the ISP view. See Changing the Model Type for more information.

Sobol Screening from Model

Opens the "Create Sobol Screening" window.

Error over Model Size (Test Data)

Error over Model Size (Training Data)

Opens the "Error on Traing-/Testdata over Model Size" window.