Hi Mehmed,
my personal experience is that only using correlation coefficients can be a great start for sure, but often it’s just not sufficient:
(Source)
What I mean with it: correlation just captures the linear dependency between features, but you can also have highly structured patterns that are not indicated by just looking at the Pearson correlation coefficient (e.g. when they are highly non-linear):
- e.g take this example: https://global.discourse-cdn.com/dlai/original/3X/7/0/707465c852ad15d58062be5913490548e8765ac3.png
- e.g. take a look at this thread: Anomaly Detection Algorithm Statistical Independence - #3 by Christian_Simonis
Hope that helps, @mehmet_baki_deniz!
Best regards
Christian