F1スコア

F1スコア

F1スコアは、モデルのパフォーマンスを評価するために使用される統計的指標です。 classification model, particularly in scenarios where the class distribution is imbalanced. It serves as a balance between precision and recall, providing a single metric that captures both false positives and false negatives.

精度は、真陽性予測の割合を予測陽性の総数で割ったもので、予測された陽性ケースのうち実際に正しかった割合を示します。一方、リコールは、真陽性予測の割合を実際の陽性の総数で割ったもので、実際の陽性ケースのうちモデルが正しく識別した割合を測定します。

F1スコアは次の式を用いて計算されます：

F1 = 2 * (Precision * Recall) / (Precision + Recall)

この式は、F1スコアが 調和平均 of precision and recall, which means it tends to be lower when either precision or recall is low. This property makes it particularly useful in situations where one metric cannot be sacrificed for the other, ensuring that both the accuracy of positive predictions and the model’s ability to capture actual positive instances are taken into account.

The F1 Score ranges from 0 to 1, where 1 indicates perfect precision and recall (all positive predictions are correct, and all positive cases are identified), while a score of 0 indicates the worst performance. It is widely used in fields such as 自然言語処理, medical diagnosis, and fraud detection, where the cost of false positives and false negatives can significantly impact outcomes.