Skip to main content

Table 4 Recap of the relationship between the multi-category metrics and the basic rates of the confusion matrix

From: The Matthews correlation coefficient (MCC) is more reliable than balanced accuracy, bookmaker informedness, and markedness in two-class confusion matrix evaluation

Scenario

Basic rates condition

# guaranteed high basic rates

high MCC means:

high TPR, TNR, PPV, and NPV

4

high BA means:

high TPR, TNR, and at least one of PPV and NPV

3

high BM means:

high TPR, TNR, and at least one of PPV and NPV

3

high MK means:

high PPV, NPV, and at least one of TPR and TNR

3

high F1 score means:

high PPV and TPR

2

high accuracy means:

high TPR and PPV, or high TNR and NPV

2

  1. #: integer number. MCC: Matthews correlation coefficient (Eq. 5). BA: balanced accuracy (Eq. 6). BM: bookmaker informedness (Eq. 7). MK: markedness (Eq. 8). F1 score: harmonic mean of TPR and PPV (Supplementary information). Accuracy: ratio between correctly predicted data instances and all data instances (Supplementary information). We call “basic rates” these four indicators: TPR, TNR, PPV, and NPV
Back to article page

BioData Mining

ISSN: 1756-0381