Explain model metrics in ML API

[ybressler]

When accessing the model prediction from the ML API, it would be helpful to have an explanation of each metric.

Possible solutions:

• Include in model prediction. (Make an additional key in json root.)
• An additional endpoint in model fit? (This would avoid redundant communication between applications.)

[ybressler]

If you're fitting data over and over again, you don't want all that extra information sitting in your json response. Plus, that content would be updated manually and should be kept away from the automated metrics of the model.