Hi all I have been using treebagger for classification of image features. Also i used prediction method from Treebagger class for predicting response after training my random forest. For the same I have following question::
1) Am i required to give 'normalized data' to treebagger for training or I can input data to treebagger without normalization? I tried both, apparently prediction is way better when i train random forest with UN-Normalized data... while out of bag error remains the same.. With this am very much confused with the inbuilt Treebagger function..
Can anyone explain me what would be the ideal way to train?(Input- Normalized or not normalized data)