Trainable Models¶

A trainable model defines the features used in training, how those features map to IFS Cloud data, and the configuration sent to the ML Service. This page lists all Trainable Models. For more information on a single model, see the Trainable Model page.

Training Status shows the progress of the latest training request:

Untrained - Initial state of the model.
Posted - The background job to start training is posted in IFS Cloud.
Preparing - The background job to start training is running. The training data to send to the training pipeline is collected at this stage.
Training - The training pipeline is being executed.
Trained - The training pipeline completed successfully.
Trained with Warnings - The training pipeline completed with warnings.
Error - The training pipeline stopped with an error.

Model Status shows the state of the model:

Inactive - Initial status of the model.
Activating - The model is being activated.
Active - The model is available to be used.

Latest Model indicates if the latest model is activated.

Train Model¶

This sends the features, configuration, and training data for the selected Trainable Model to the ML Service. It is available when the training status is Untrained, Error, Trained, or Trained with Warnings.

Activate Model¶

This activates a trained model that is not the latest model. The model must be activated before it can be used for predictions. It is available when the training status is Trained or Trained with Warnings, the model status is not Activating, and the model is not already the latest model.

Refresh Training Status¶

This queries the training status of the selected Trainable Model from the ML Service and displays it on the record. It is available when the training status is Training.

The Database Task Schedule Refresh Machine Learning Models refreshes the training status of any trainable model in status Training. It runs on regular intervals, every ten minutes by default, and can be changed on the Database Task Schedule page by using View Schedule.

Refresh Model Status¶

This queries the model status of the selected Trainable Model from the ML Service and displays it on the record. It is available when the training status is Trained or Trained with Warnings.

The Database Task Schedule Refresh Machine Learning Models refreshes the model status of any trainable model in status Activating. It runs on regular intervals, every ten minutes by default, and can be changed on the Database Task Schedule page by using View Schedule.

Reset Training Request¶

This resets a training request. It is available when the training status is Preparing or Posted.

Edit Parameters¶

This opens a dialog where the parameters for the Trainable Model can be modified. It is available when the training status is Untrained, Trained, Trained with Warnings, or Error.

View Parameters¶

This opens the same dialog in read-only mode. It is available when the training status is Posted, Preparing, or Training.

Add Schedule¶

This adds a schedule for the Trainable Model when the model does not already have one. The default schedule is weekly on Sunday at 00:00.

View Schedule¶

This opens the database task schedule for the Trainable Model. It is available when a schedule exists. The schedule can be edited and activated there. Activating the schedule re-trains the model on data from IFS Cloud at the scheduled time. The Next Execution Date is not displayed unless the schedule is activated.

View Background Jobs¶

This allows viewing the history of the model training jobs that were run.

More details related to a background job can be viewed by selecting Show Details for the background job.

View Application Messages¶

This allows viewing the message requests and responses to and from the ML Service.

More details related to an application message can be viewed by selecting View Details for the application message.

View Logs¶

This will navigate to the Machine Learning Logs page and display logs for the selected Trainable Model.