Taking into account lot of techniques for features selection, how to select the best algorithm without test all of them ? What Methodology must carry out after literature review based on type of data ?
I'd start here:
https://web.stanford.edu/~hastie/StatLearnSparsity/
This book is based mainly on papers published elsewhere, so if you have specific needs you might be alright with individual papers. I prefer the book form as it makes clear the relationships among the papers. Tibshirani and colleagues have other reviews of the techniques too. I should add, this is feature selection and shrinkage, but there are good reasons for many applications not to use all-or-none feature selection (I usually cite Efron et al.'s lars paper for this).
Thanks for your answer the book is very good. However, the question Is what do methodology must apply for selecting Thanks for your answer the book is very good. However, the question Is what methodology should I apply for selecting the feature selection algorithm ?
and what is the title of Efron’s paper.?
- Badges
- Science method
- Similar topics
- Methodology