article Open AccessTop 1% cited
The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation
BMC Genomics · 2020 · Vol. 21(1) · pp. 6–6
Davide Chicco✉(Krembil Foundation)Giuseppe Jurman(Fondazione Bruno Kessler)
Abstract
In this article, we show how MCC produces a more informative and truthful score in evaluating binary classifications than accuracy and F<sub>1</sub> score, by first explaining the mathematical properties, and then the asset of MCC in six synthetic use cases and in a real genomics scenario. We believe that the Matthews correlation coefficient should be preferred to accuracy and F<sub>1</sub> score in evaluating binary classification tasks by all scientific communities.
Imbalanced Data Classification TechniquesText and Document Classification TechnologiesData Mining Algorithms and ApplicationsBinary classificationFalse positive paradoxBinary numberFalse positives and false negativesCorrelationConfusion matrixArtificial intelligenceStatisticsPearson product-moment correlation coefficientFalse positive rate
MeSH terms
Machine LearningCorrelation of DataAlgorithmsData Interpretation, StatisticalComputational Biology
Funding
- University of Toronto
Citations
5,520
FWCI
407.89
field-weighted impact
References
104
Percentile
100%
vs. same field & year
Citations per year
References
The Precision-Recall Plot Is More Informative than the ROC Plot When Evaluating Binary Classifiers on Imbalanced Datasets
PLoS ONE · 2015 · 4,310 citations
Measures of the Amount of Ecologic Association Between Species
Ecology · 1945 · 11,764 citations
Broad patterns of gene expression revealed by clustering analysis of tumor and normal colon tissues probed by oligonucleotide arrays
Proceedings of the National Academy of Sciences · 1999 · 4,201 citations
Morphometric analysis of white matter lesions in MR images: method and validation
IEEE Transactions on Medical Imaging · 1994 · 1,264 citations
Measuring classifier performance: a coherent alternative to the area under the ROC curve
Machine Learning · 2009 · 976 citations
The use of the area under the ROC curve in the evaluation of machine learning algorithms
Pattern Recognition · 1997 · 7,122 citations
The meaning and use of the area under a receiver operating characteristic (ROC) curve.
Radiology · 1982 · 21,496 citations
The Measurement of Observer Agreement for Categorical Data
Biometrics · 1977 · 77,212 citations
Related articles
The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation
BMC Genomics · 2020 · 5,520 citations
Citation Network
How this paper connects to the literature. Drag to explore, click any node to open that paper.