If we shuffle one feature , for a particular example the actual value for the shuffled feature is wrong and other features values are correct. Does it have any effect on this because of the wrong value?
What formula is used to calculate the feature-derived error? This is an interesting new idea as in the past when I wanted to assess the importance of features, I used simple linear correlation between all features and the target variable. And this could potentially have issues, although I couldn't think about other solutions.
what is the difference between Recursive Feature Engineering vs Shuffle Feature Importance ?
I know RFE we should give number of features to be selected !! , is their any different way both happens ?
If we shuffle one feature , for a particular example the actual value for the shuffled feature is wrong and other features values are correct. Does it have any effect on this because of the wrong value?
What formula is used to calculate the feature-derived error? This is an interesting new idea as in the past when I wanted to assess the importance of features, I used simple linear correlation between all features and the target variable. And this could potentially have issues, although I couldn't think about other solutions.