Slightly lower overall accuracy might be the best classifier for your goal. The best overall score might not be the best model for your goal. Not set aside for testing, counting each observation when it was in aįor holdout validation, the score is the accuracy on the held-outįor resubstitution validation, the score is the resubstitutionĪccuracy against all the training data observations. For cross-validation, the score is the accuracy on all observations
