3

I am trying to calculate precision for multi class classification for different datasets. but mostly I am getting precision value as 1. what does it mean? any wrong with my algorithm?

padman mukharjee
  • 31

1 Answers1

4

This means that all your positive samples are classified as positive samples and none of the positive samples are classified incorrectly. Precision alone will not tell you enough about the classifiers performance. Better if you also look at the accuracy (if your data isn't imbalanced), recall, f1-score and if possible ROC AUC score. For example you can have a high precision, but still have a low accuracy, recall or f1-score

Olivier_s_j
  • 1,185
  • 2
    An example why precision is not enough is that you can classify everything as positive and you have a nice fine 1 value for precision and an useless classifier. – rapaio May 08 '14 at 09:00