Extending F1 metric, probabilistic approach
This work provides an incremental improvement to evaluation metrics for binary classification in machine learning.
The authors tackled the problem of assessing binary classifier performance by proposing a new probabilistic extension of the F1 score, which they tested on a real dataset to demonstrate its properties.
This article explores the extension of well-known F1 score used for assessing the performance of binary classifiers. We propose the new metric using probabilistic interpretation of precision, recall, specificity, and negative predictive value. We describe its properties and compare it to common metrics. Then we demonstrate its behavior in edge cases of the confusion matrix. Finally, the properties of the metric are tested on binary classifier trained on the real dataset.