Model creation is an iterative process, depending on the outcome of an iteration, the user might change model parameters (which learners, confidence settings etc) or data filters to optimize the result. Right now entering the model parameters has to be done from scratch every time. The suggestion is to store model parameterization and filter, similar to the "Input Set" in the Thingworx Service Editor, so it can be reused. When a user creates a new model, he could then select a saved configuration to quickly take over the values from a previous model creation.
As an alternative, instead of storing model configuration as a new object, in model creation, the user could just select an existing model as "parameter template" to take over its configuration.