A rec-center field-permit helper reports 94% “correct intent.” Almost all failures are names from one language community. What measurement step is required?
Select an answer to reveal the explanation.
Short Explanation
94% correct intent, and almost all the misses are names from one language community. Review performance by subgroup, not only the headline number. Publishing 94%, swapping to a generic F1, or dropping that community from the test set hides the gap.
Full Explanation
Minimizing bias includes checking performance across groups, not only a headline average. A 94% intent score can hide a community that almost always fails. A generic experiment metric or dropping the missed group conceals the problem. Subgroup review is the required measurement step.