Accuracy
The fraction of predictions that were correct.
What people get wrong
Meaningless on imbalanced data. If one class is 1% of the rows, a model that always predicts the other class scores 99% and detects nothing.
A definition is not an understanding. This is taught properly in Choosing the metric that matches the cost of being wrong, part of Model evaluation & metrics.
Learn this properlyTerms from the same topic