Note 1 : Recommended statistics for this type of classification highlighted in aqua
Note 2 : The recommendation system assumes the input is the result of classification over the entire dataset, not just a subset. If the confusion matrix is based on test data classification, the recommendation may not be valid.
Actual | Predict
|
Class | L1 | L2 | L3 | Description |
ACC | 0.83333 | 0.75 | 0.58333 | Accuracy |
AGF | 0.72859 | 0.62869 | 0.61009 | Adjusted F-score |
AGM | 0.85764 | 0.70861 | 0.58034 | Adjusted geometric mean |
AM | -2 | 1 | 1 | Difference between automatic and manual classification |
AUC | 0.8 | 0.65 | 0.58571 | Area under the ROC curve |
AUCI | Very Good | Fair | Poor | AUC value interpretation |
AUPR | 0.8 | 0.41667 | 0.55 | Area under the PR curve |
BB | 0.6 | 0.33333 | 0.5 | Braun-Blanquet similarity |
BCD | 0.08333 | 0.04167 | 0.04167 | Bray-Curtis dissimilarity |
BM | 0.6 | 0.3 | 0.17143 | Informedness or bookmaker informedness |
CEN | 0.25 | 0.49658 | 0.60442 | Confusion entropy |
DOR | None | 4.0 | 2.0 | Diagnostic odds ratio |
DP | None | 0.33193 | 0.16597 | Discriminant power |
DPI | None | Poor | Poor | Discriminant power interpretation |
ERR | 0.16667 | 0.25 | 0.41667 | Error rate |
F0.5 | 0.88235 | 0.35714 | 0.51724 | F0.5 score |
F1 | 0.75 | 0.4 | 0.54545 | F1 score - harmonic mean of precision and sensitivity |
F2 | 0.65217 | 0.45455 | 0.57692 | F2 score |
FDR | 0.0 | 0.66667 | 0.5 | False discovery rate |
FN | 2 | 1 | 2 | False negative/miss/type 2 error |
FNR | 0.4 | 0.5 | 0.4 | Miss rate or false negative rate |
FOR | 0.22222 | 0.11111 | 0.33333 | False omission rate |
FP | 0 | 2 | 3 | False positive/type 1 error/false alarm |
FPR | 0.0 | 0.2 | 0.42857 | Fall-out or false positive rate |
G | 0.7746 | 0.40825 | 0.54772 | G-measure geometric mean of precision and sensitivity |
GI | 0.6 | 0.3 | 0.17143 | Gini index |
GM | 0.7746 | 0.63246 | 0.58554 | G-mean geometric mean of specificity and sensitivity |
HD | 2 | 3 | 5 | Hamming distance |
IBA | 0.36 | 0.28 | 0.35265 | Index of balanced accuracy |
ICSI | 0.6 | -0.16667 | 0.1 | Individual classification success index |
IS | 1.26303 | 1.0 | 0.26303 | Information score |
J | 0.6 | 0.25 | 0.375 | Jaccard index |
LS | 2.4 | 2.0 | 1.2 | Lift score |
MCC | 0.68313 | 0.2582 | 0.16903 | Matthews correlation coefficient |
MCCI | Moderate | Negligible | Negligible | Matthews correlation coefficient interpretation |
MCEN | 0.26439 | 0.5 | 0.6875 | Modified confusion entropy |
MK | 0.77778 | 0.22222 | 0.16667 | Markedness |
N | 7 | 10 | 7 | Condition negative |
NLR | 0.4 | 0.625 | 0.7 | Negative likelihood ratio |
NLRI | Poor | Negligible | Negligible | Negative likelihood ratio interpretation |
NPV | 0.77778 | 0.88889 | 0.66667 | Negative predictive value |
OC | 1.0 | 0.5 | 0.6 | Overlap coefficient |
OOC | 0.7746 | 0.40825 | 0.54772 | Otsuka-Ochiai coefficient |
OP | 0.58333 | 0.51923 | 0.55894 | Optimized precision |
P | 5 | 2 | 5 | Condition positive or support |
PLR | None | 2.5 | 1.4 | Positive likelihood ratio |
PLRI | None | Poor | Poor | Positive likelihood ratio interpretation |
POP | 12 | 12 | 12 | Population |
PPV | 1.0 | 0.33333 | 0.5 | Precision or positive predictive value |
PRE | 0.41667 | 0.16667 | 0.41667 | Prevalence |
Q | None | 0.6 | 0.33333 | Yule Q - coefficient of colligation |
QI | None | Moderate | Weak | Yule Q interpretation |
RACC | 0.10417 | 0.04167 | 0.20833 | Random accuracy |
RACCU | 0.11111 | 0.0434 | 0.21007 | Random accuracy unbiased |
TN | 7 | 8 | 4 | True negative/correct rejection |
TNR | 1.0 | 0.8 | 0.57143 | Specificity or true negative rate |
TON | 9 | 9 | 6 | Test outcome negative |
TOP | 3 | 3 | 6 | Test outcome positive |
TP | 3 | 1 | 3 | True positive/hit |
TPR | 0.6 | 0.5 | 0.6 | Sensitivity, recall, hit rate, or true positive rate |
Y | 0.6 | 0.3 | 0.17143 | Youden index |
dInd | 0.4 | 0.53852 | 0.58624 | Distance index |
sInd | 0.71716 | 0.61921 | 0.58547 | Similarity index |
Generated By PyCM Version 4.1