Machine learning on or of some data?

If it’s machine learning using some platform or technology, it makes sense to use "on" ("ML on AWS"), but if we are talking about applying ML to some data, should "on" or "of" be used? Below two examples of real world usage:

• …machine learning of collective variables…
• …machine learning on binary data…

My specific question is about the phrase "Using machine learning of demographics and personal history to predict XYZ." Should that be "Using machine learning on demographics and personal history to predict XYZ" instead?

There might be a better Stack Exchange site to post this on, but I wasn’t sure. Feel free to move or let me know if this is the case.