The number of known-defective units missed by an inspection divided by the known-defective units checked.
Also called: false negative rate; missed detection rate.
What is the denominator of miss rate?
With positive defined as the target defect being present, a missed defect is a false negative: the unit is defective but the inspection accepts it. Miss rate, or FNR, divides these misses by all known-defective samples checked for that defect.
Keep the observation unit explicit: a whole bottle, a tablet event and a defect image are different samples. Naming the target defect and reference group is necessary before comparing two percentages.
Why can the same misses produce two different percentages?
Dividing misses by all inspected samples gives a different proportion from FNR, whose denominator contains only known-defective samples. Two reports can therefore assign different percentages to the same missed units.
Does zero observed misses prove zero miss risk?
No. With known-defective samples present, zero observed misses is a result for those samples and conditions. It is not a guarantee for untested presentations or future production. With no known-defective samples, the FNR denominator is zero and this rate cannot be estimated from the dataset.
For the same positive class and reference-labelled dataset, FNR equals one minus recall. Physical removal of flagged bottles is a later function assessed through reject confirmation.