0

I have a scenario of multi-class dataset with around 10 distinct classes of target. There are 3 categorical features each with multiple labels. If we check the data, each unique combination of feature values will always return single class of Target. So any classifier model should not get confused what would be possible target for any given combination of features.So I'm anticipating 100% accuracy. But none of the classifier models able to return 100% accuracy. I tried Random forest, xgboost. Note, features are converted by get_dummies. Label is converted by labelenconder. Applied random SMOTE for equal distribution. Can anyone suggest if I'm missing anything to consider?

soumalya saha
  • 1
  • 2
  • 1
    Why did you apply SMOTE? – Dave Jan 04 '23 at 15:43
  • Hi,I applied SMOTE as distribution across different classes were not equal. Basically to increase the number of cases where there was not enough records available. SMOTE indeed helped to increase the accuracy. – soumalya saha Jan 06 '23 at 03:28

0 Answers0