0
$\begingroup$

I have a dataset of 45 non-linear numerical values and 2 categorical values. I am making a feature selection to predict categorical variables one by one or together. I used the correlation ratio and kendall rank correlation coefficient to calculate the strenght of the relations.

Among these two methods, which one should I use as the primary method to sort the strenght of variables? For example the table below is sorted by correlation ratio. Should I sort the table with the kendall coefficient instead?

correlation table

$\endgroup$
  • $\begingroup$ did you try Lasso/Ridge? $\endgroup$ – Peter May 13 at 22:46
  • $\begingroup$ Actually I just looked at them. Linear, Lasso and Ridge regression are used for feature selection, but I am not clear on which one to choose? My data is non-linear so I think Linear regression won't be an option. Among the other two, which one is more suitable if I want to determine my case with 45 non-linear numerical predictors and 2 categorical targets? $\endgroup$ – Jean May 14 at 15:11

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.