Machine learning model lifecycle

The InfoSphere® MDM machine learning assisted data stewardship service uses models that determine the service's decision making guidance. There are several stages in each model's lifecycle, starting from the moment it starts to be trained.

A machine learning assisted data stewardship model goes through the following stages in its lifecycle:

Training

Models with a status of training are in the process of being trained using the latest training data. When a model has been trained, it is put into the inactive state. You can use the REST API's train endpoint to begin training a new model.

Inactive

An inactive model status means that the model is fully trained and ready to be made active at any time. If an active model is deactivated, then it is put back into the inactive state.

Active

When a model is the current, active model in the system, it is in the active state. You can use the REST API's activate endpoint to put a model into an active state.

Error

There is an error associated with this model. Errors normally occur during the training phase of a model's lifecycle. Delete the model, address the cause of the error, and then restart the training.

Deleted

The model has been deleted from the system. You can delete a model using the REST API's delete model_id endpoint.