No Number of Neighbors fit to SMOTE Upsampling

aylumnaaylumna Member Posts: 3 Contributor I
I use SMOTE to handle my imbalance class (the class contains 3 types). But it shows error every i run the process. The error shows "Too many neighbors". Even i set to the number of neighbors is 1,2,3,4 or 5.
Tagged:

Answers

  • ceaperezceaperez Member Posts: 333 Unicorn
    Hello @aylumna
    Probably this is due to your dataset doesn't contain enough samples of your minority class. Check the attribute and the observations per class. 

    Best
    Cesar
Sign In or Register to comment.