Models lifecycleΒΆ

Using machine learning in DSS is a process in two steps:

  • The models are designed, trained, explored and selected in the Lab

  • Once you are satisfied with your model, you Deploy it from the lab to the Flow, where it appears as a Saved model

A Saved model is deployed together with a Training recipe that allows you to retrain the saved models, with the same model settings, but possibly with new training data.

A Saved Model on the Flow can be used:

  • For real-time APIs, using the API node

  • By a Scoring recipe, in order to perform prediction (or clustering) on a non-labelled dataset