Straightforward alternative "single number" evaluation metric for ML algorithms--- should make comparisons to e.g. Rea et al (2019) and Churchill (2019) a bit easier.
More generally, we could output various Fγ scores, such as the F0.5, F1, F2 measures.
Straightforward alternative "single number" evaluation metric for ML algorithms--- should make comparisons to e.g. Rea et al (2019) and Churchill (2019) a bit easier.
More generally, we could output various Fγ scores, such as the F0.5, F1, F2 measures.
See eq. 2 in Rea et al (2019), van Rijsbergen (1979) pg 134, Sasaki (2007), or https://en.wikipedia.org/wiki/F1_score for example.