Water-plant 'overflow event' rows are about one percent of the week. A model that always predicts no overflow posts 99 percent accuracy and is useless. What is wrong with that headline metric?
Select an answer to reveal the explanation.
Short Explanation
Overflow rows are one percent of the week. A dummy that always says no overflow posts 99 percent accuracy and catches nothing. Accuracy is a trap on a rare class; pick a metric that moves with the event.
Full Explanation
On a rare-event classifier, a majority-class dummy can post high accuracy and still miss every event. Reject accuracy as the headline and use a metric that moves with the rare class. Polly is irrelevant, and this is evaluation, not Domain 1 resampling.