This processor is used to load a trained model generated by other processors, and apply this model on an input Dataset to generate prediction/forecast.
The Model Application processor needs two inputs:
- Trained Model: Presents the model to load in order to make a prediction/forecast and it can be selected within the processor configuration
- Dataset to forecast: This dataset must match in the schema with the dataset on which the model was trained (column names and types) and it is provided via the input node of the processor
- if at least one wrong column name is provided, a Processor Execution Error is generated
- if at least one wrong column type is provided, a Processor Execution Error is generated
The configuration menu of the processor looks as follows:
NOTE THAT: the user can only select owned/shared models
This processor uses the selected model to apply prediction/forecast on the input Dataset, and it gives as output this Dataset with an additional column containing the prediction/forecast.
NOTE THAT: the prediction column name is provided by the model (it is given in the configuration of the model within the processor that generates it)
In this example, the Model Application Processor will be applied to generate predictions for an input Dataset generated by a Custom Input Table. This dataset contains some test entries of the famous Iris Dataset
The used model is a Machine Learning Model trained by a Train Model Processor.
The Workflow which generated this model is as follows:
Train Model Processor