approximated generalized likelihood ratio(English)
- rejects H0 when nẌ2 > χ21 (α), the upper-α point of the χ1 2 distribution. This is the same as rejecting when |X¯| > z(α/2)/√n, so the GLRT is equivalent to usual two-sided z-test based on X
- ACC, AURC, REML, RMS, RMT, RoR, SMA, TMS
- Statistics, Ratio-based metric
- https://doi.org…nhum.2016.00204
- https://web.sta…2/Lecture22.pdf