Many online posts(here and here) say label encoding would make NN/DL models think there is an ordering among categories, so it should not be used for categorical encoding for NN/DL models. But wouldn't NN/DL models be capable of modeling any non-linear relationship like XGboost? If this is the case, why NN/DL can't figure out whether or not an ordering is actually present among categories when label encoding is used?
Asked
Active
Viewed 18 times
0
-
1Could you share some context where it was said? – Tim Nov 02 '20 at 20:10
-
@Tim examples are added. – etang Nov 02 '20 at 20:27
-
The links seem to answer your question, don't they? Nonetheless, this seems to be already answered on our site, so check the linked thread. – Tim Nov 02 '20 at 20:33
-
@Tim I disagreed with them and offer the reason(NN modeling non-linear relationship) for asking it again. – etang Nov 02 '20 at 21:30
-
What exactly you disagree with and why does the linked thread not answer your question? – Tim Nov 02 '20 at 21:44