Skip to main content
. 2021 Feb 17;3:602683. doi: 10.3389/fdgth.2021.602683

Table 7.

Performance comparison of multilabel and binary classification setups for each offense type (best results are highlighted in green).

Hands-on Hands-off ADVO breach
Setup Accuracy ROC F1 Accuracy ROC F1 Accuracy ROC F1
Multilabel (50%) 0.6520 0.6493 0.7400 0.6043 0.5753 0.7300 0.6900 0.6676 0.8000
Multilabel 0.6587 0.6543 0.7600 0.6065 0.5786 0.7300 0.6840 0.6540 0.8000
Binary 0.6575 0.6456 0.7578 0.6072 0.5749 0.7367 0.6882 0.6576 0.8012