Thereis a notion presented a few years ago that continues to prompt discussion.This notion is that given enough statistical evidence, it's no longer necessary to understand why things happen - we need only know what things happen together.

Discuss the circumstances when it makes sense to act on a correlation and when one should not act.

Many data science models used for making predictions are based on correlated variables.

How should such models be developed and used so the predictions can be acted upon?

Some of models are very complex.Do we need to understand why they work?