Hi everyone, Just a more general question. I didn’t quite understood why different metrics have been used for comparing the feature selection approaches? Why not simply use the precision and recall along with the number of features selected?
Thank you very much for your question. Given the problem you are trying to solve, different accuracy measurements can be used to help you better evaluate your model. For example, if you are trying to avoid false positives or false negatives you may consider using precision or recall. On the other hand, if you are dealing with unbalanced datasets, you might want to consider using true positive or negative rates as they are not influenced by the number of datapoints in different classes. Using this link you can read about different measurements and see when/how you can use them.
All the best,