I have 315 data with 12076, I need to reduce the dimension and also select just 10 most important features. What is the best way? Is it better to do feature extraction and choose the features with high weights or do a feature selection method or any other method? Is it a good way to first do feature selection to reduce the dimension and then feature extraction to find the best space and rotation for my data?

  • $\begingroup$ If you have a response variable that you want to predict then PLS(Partial Least Square) is the way to go else Principal Components Analysis will help you to reduce the dimensions. $\endgroup$ – ahmedrajput Jul 2 '17 at 19:22
  • $\begingroup$ @ahmedrajput I know for example PCA doesn't consider the label of data, but the problem is, which one is better. If I reduce the dimension and choose features with most weights or use feature selection directly. $\endgroup$ – user137927 Jul 3 '17 at 21:17

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.