What are false negatives?
False negatives are the cases a classifier labels as negative when they are actually positive. A maintenance model that reports a gearbox as healthy the week before it seizes, a screening test that clears a patient who has the disease, and a fraud model that approves a stolen card have all produced false negatives. In statistics the mistake is called a Type II error. False negatives are the misses that recall measures: recall equals true positives divided by true positives plus false negatives, so every false negative pulls recall down.
Why false negatives matter
False negatives are the silent cost of a model. A false positive announces itself when someone investigates and finds nothing. A false negative produces no alert at all, so the failure arrives later as an unplanned outage, a chargeback, or a missed diagnosis. That makes false negatives easy to under-count unless the team deliberately looks for them.
A paper mill runs a model that predicts bearing failures on its dryer section from vibration data. In its first year the model flagged many bearings early, but two failures happened with no warning. Both bearings turned out to be a newer type with a different vibration signature that was missing from the training data. Counting those false negatives led to a retraining with the new type included. Lowering the confidence threshold catches more positives but adds false positives, so the balance must be set with the people who pay for both kinds of error.