I built a decision tree and oversampled the minority class using smote. After this, class 2 (from classes 0, 1, 2, 3) is being completely ignored (for the unbalanced test set). Nothing is classified as class 2 correctly or wrong. How can this be?
Asked
Active
Viewed 37 times
1 Answers
0
It is not possible. Smote is basically used to balance the classes. All the classes i.e., 0,1,2,3 will have equal number of rows after using smote.
Yashi Aggarwal
- 407
- 2
- 6
-
The classes are balanced now, but the result with the test set is even more unbalanced than without smote – maybeyourneighour Aug 20 '19 at 10:19