Table 2 Performance measures for multi-class classifiers. t p i , t n i , f p i and f n i are, in turn, tp, tn, fp and fn for class i [59]. The multi-class MCC is calculated taking into account all the entries of the confusion matrix C K×K involving all K classes [60]. The ij-th entry (c ij ) is the number of examples of the true class i that have been assigned to the class j by the classifier
Measure | Formula | |
|---|---|---|
Accuracy | \(\frac {\sum _{i=1}^{K}\frac {tp_{i}+tn_{i}}{tp_{i}+fn_{i}+fp_{i}+tn_{i}}}{K}\) | |
MCC | \(\frac {\sum _{k,l,m=1}^{K}C_{\textit {kk}}C_{\textit {ml}} - C_{\textit {lk}}C_{\textit {km}}}{\sqrt {\sum _{k=1}^{K}\left [\left (\sum _{l=1}^{K}C_{\textit {lk}}\right)\left (\sum _{f,g=1 f\not =k}^{K}C_{\textit {gf}}\right)\right ]}\sqrt {\sum _{k=1}^{K}\left [\left (\sum _{l=1}^{K}C_{\textit {kl}}\right)\left (\sum _{f,g=1 f\not =k}^{K}C_{\textit {fg}}\right)\right ]}} \) |