| Data Modeler Op |
|
The Data Modeler Op is an operator which can apply a data prediction model on a product. A prediction model can be created beforehand with the Data Prediction Tool. The outcome of the operator as well as some of the parameters depend largely on the selected model.
The Data Modeler Op can be invoked from the VISAT tool menu by selecting the Data Model Application... command.
On the command line the processor is available by means of the Graph Processing Tool gpt
which is located in the BEAM bin directory.
Typing gpt Data.Modeler.Op -h displays further information.
Selecting the Data Modeler Op command from the VISAT tool menu pops up the following dialog:
Name: Here the user specifies the source product. The combo box presents a list of all products open in VISAT. The user may select one of these or, by clicking on the button next to the combo box, choose a product from the file system.
Name: Used to specify the name of the target product.
Save as: Used to specify whether the target product should be saved to the file system. The combo box presents a list of available file formats. The text field or the button next to it allow to specify a target directory.
Open in VISAT: Used to specify whether the target product should be opened in VISAT. When the target product is not saved, it is opened in VISAT automatically.
When the Processing Parameters tab is opened, it will look like this:
Here the model-independent parameters are listed.
Model file: Here you can choose a model from the file system by clicking on the button next to the text field. Only models which are compatible with the selected source product can be imported.
Output source product bands: If checked, the bands from the source product will be copied to the target product.
Band suffix: Here you can specify a suffix which will be appended to each band created by the model. Also, a group will be created which contains the bands.
Use valid pixel expression: If checked, the expression stated in the text field below will be considered during processing. The computation will only be performed for pixels meeting the expression.
If you choose a model, the tab will display more parameters. The image below shows parameters for a Maximum Likelihood model.
Output Bands: Here all bands which can be created by the model are listed. The user can check which of these bands should be written to the target product.
Correspondence between model dimensions and bands: The user can assign bands or the source product to the source dimensions in the model. This might be necessary if the band names in the products are different. If a band name appears in both products, the assignment will be performed automatically.
Model Application Parameters: Here method-specific parameters are listed. If a method does not offer any parameters, this will be empty. In the example above, the Maximum Likelihood Classifier offers to set a distinct probability threshold for each class.
Run Creates the target product. The processing is actually deferred until its band data are accessed, either by writing the product to a file or by viewing its band data. When the Save as option is checked, the processing is triggered automatically.
Close Closes the dialog.
Help Displays this help page.
File The entries found in this menu allow to save the current processing parameters to disk and to open a saved parameter set from disk.
Help Displays this help page.