Hypothesis testing
Hypothesis testing: A statistical procedure comparing data with a null hypothesis to assess evidence against a reference model. It requires metrics and hypotheses defined before analysis. Multiple testing increases false positives and calls for corrections or preregistered plans. This perspective connects the definition with practical behaviour in artificial intelligence systems.
In italiano: Test di IpotesiLegge V · Overfitting, bias e collasso dei modelliIt requires metrics and hypotheses defined before analysis. Multiple testing increases false positives and calls for corrections or preregistered plans.