Statistics and Machine Learning use data as numbers or symbols.
But these way has limits to use the data.
If we know the backgrounds or system of the data, the use of the data becomes wider.
Easier way is to use the methods of Change data for Analysis.
The other is that we use the data physics in the model building. I think 3 methods below are uses the idea of data physics.
In data physics, local idea is useful.
NEXT Data of Physics