Model evaluation / validation tools

[evangeline-corcoran]

Potential users in domains such as ecology/biology have identified a lack of clear ways to evaluate the results of models and compare different models for their use case as a barrier for application of machine learning to data from these fields.

It would be good to discuss:

• To what extent could be provide this in scivision?
• Should users be able to specify appropriate evaluation methods for their model when adding it to the catalogue?