Honestly, I think ML should always involve at least a little bit of coding, which would be more practical. That said, this looks reasonable, good playground for experiment.

A good similar product is Orange: https://orangedatamining.com/