Accuracy
An evaluation metric for classification models that computes the ratio of correct predictions to total predictions. In IDP scenarios with many blank fields, it tends to be inflated because correctly identified blank fields (true negatives) are counted.
Core Facts
When samples are severely imbalanced, Accuracy becomes misleading—a model that predicts everyone as healthy can still achieve an accuracy of over 99%.
The formula for Accuracy is (TP+TN)/(P+N), i.e., the proportion of correctly predicted samples out of the total number of samples.
Timeline (last 90 days)
In form scenarios with many blank fields, if correctly identifying blank fields (correct blank) is counted as a True Negative, Accuracy becomes inflated, potentially masking the model's shortcomings in real information extraction.
The Jaccard Index (IoU) is described as "Accuracy without True Negatives," focusing only on positive overlap and not involving TN counts, making it more robust than Accuracy in form scenarios with many blank fields.
All Facts (4)
When samples are severely imbalanced, Accuracy becomes misleading—a model that predicts everyone as healthy can still achieve an accuracy of over 99%.
70%VerifiedThe formula for Accuracy is (TP+TN)/(P+N), i.e., the proportion of correctly predicted samples out of the total number of samples.
65%UnverifiedIn form scenarios with many blank fields, if correctly identifying blank fields (correct blank) is counted as a True Negative, Accuracy becomes inflated, potentially masking the model's shortcomings in real information extraction.
50%UnverifiedThe Jaccard Index (IoU) is described as "Accuracy without True Negatives," focusing only on positive overlap and not involving TN counts, making it more robust than Accuracy in form scenarios with many blank fields.
50%