I am very confused: For what I understood I should:

  1. Multicollinearity check with Pearson corr and possibly consider to drop multicolliner features Then? I am very confused feature selection should be done with respect to the target variable or not and secondly in all the combinations of stat tests possible based on independent and dependent features: num cols and num target, num cols and cat target, cat cols and num target, cat cols and cat target. Sorry if the question is quite confused, but it reflects my confusion on all the possible statistical tests.
  • $\begingroup$ measuring correlation coefficient between features (eg amount of colinearity) is one of the ways to detect features which dont add new information. Another way is PCA, ICA, .. $\endgroup$ – Nikos M. Mar 28 at 12:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.