Synopsis
Applies a model to an example set. This might be a prediction or another data transformation model.
Description
This operator applies a Model to an ExampleSet. Models usually contain information about the data they have been trained on. This information can be used for predicting the value of a possibly unknown label, reproduce some transformations as during training or performing other changes. All needed parameters are stored within the model object.
Please pay attention to the fact, that the application of Models will need the same attributes during application on an ExampleSet that where part of the ExampleSet it was trained on. Some minor changes like adding attributes might be possible, but might cause severe calculation errors. Please make sure, that the attributes' number, order, type and role are consistent during training and application.
If the model supports views, it is possible to create a view instead of changing the underlying data. In order to advise the Apply Model operator to do so, simply switch on the create view parameter. The transformation that would be normaly performed directly on the data will then be computed every time a value is requested and the result is returned without changing the data. Please keep in mind, that not all models support views.
If you have to apply serveral models in a row, like for example when you have to apply a few preprocessing models before applying a prediction model, then you would like to group models. This is possible using the Group Modelsoperator in a convenient way.
Input
- model: expects: Model
- unlabelled data: expects: ExampleSetMetaData: #examples: = 0; #attributes: 0
Output
- labelled data:
- model:
Parameters
- application parameters: Model parameters for application (usually not needed).
- create view: Indicates that models should create a new view on the data where possible. Then, instead of changing the data itself, the results are calculated on the fly if needed.