Fscore

Fscore, commonly written as F-score or F-measure, is a family of metrics that combines precision and recall into a single value. In binary classification, precision = TP/(TP+FP) and recall = TP/(TP+FN). The F-score is the harmonic mean of precision and recall, for F1-score F1 = 2 * precision * recall / (precision + recall). More generally, the Fβ-score is defined as Fβ = (1+β^2) * (precision * recall) / (β^2*precision + recall), where β determines the relative weight of recall vs precision. When β > 1, recall is weighted more heavily; when β < 1, precision is weighted more heavily.

F-scores are widely used in information retrieval, NLP, and machine learning to evaluate classifiers, particularly when datasets are imbalanced. They can be extended to multiclass settings via macro, micro, or weighted averaging.

Limitations: F-score condenses two potentially separate aspects into one number, possibly hiding imbalances between precision and recall; it depends on the chosen threshold; it does not account for costs of errors or class distribution; it does not measure calibration.

Piotroski F-Score: In finance, the Piotroski F-Score is a nine-point signal-based score used to assess a firm's financial strength, incorporating profitability, leverage and liquidity, and operating efficiency. Each of nine signals contributes one point, for a total score between 0 and 9. It is named after Joseph Piotroski and is used by investors to identify firms with high earnings quality and potential for higher returns.