The most robust metric for evaluating classification models, especially on imbalanced datasets.
Confusion Matrix
Model Performance
MCC Score0.616
Superior Predictor
Accuracy80.0%
F1-Score0.800
Precision88.9%
Recall72.7%
Interpretation
+1.0 = Perfect, 0.0 = Random, -1.0 = Inverse.
Overview
The Matthews Correlation Coefficient (MCC) is widely considered the best single metric for summarizing a classification model. Unlike accuracy, it considers all four values of the confusion matrix equally, making it robust against highly imbalanced datasets.
💡
Pro Tips
Use MCC instead of Accuracy when classes are imbalanced (e.g., fraud detection).
An MCC score above 0.7 is generally considered a very strong predictor.
A score of 0 means your model is no better than a random guess.
!
Fun Facts
"MCC is also known as the phi coefficient (φ)."
"If a model predicts every single instance as the majority class, its MCC will be 0, while its accuracy could be 99% or higher."
"MCC is essentially a correlation coefficient between the observed and predicted binary classifications."
