Accuracy
Correct predictions divided by total predictions, (TP + TN) / (TP + TN + FP + FN). On imbalanced data a model can score high accuracy while never predicting the rare class, so precision, recall or F1 are often reported instead.
Accuracy = (TP + TN) / (TP + TN + FP + FN)
Example
If 95 of 100 emails are not spam, a model that always answers "not spam" scores 95% accuracy while catching zero spam.