F1 分数与阈值选择
理解精确率、召回率与 F1 分数的权衡
参数控制
真异常数量
50
总样本数
500
当前阈值 ε
0.050
当前指标
精确率 Precision
0.000
召回率 Recall
0.000
F1 分数
0.000
最佳 F1
0.000
F1 / Precision / Recall vs 阈值
F1 分数
精确率
召回率
当前阈值
最佳阈值
混淆矩阵
True Positive (TP)
0
False Positive (FP)
0
False Negative (FN)
0
True Negative (TN)
0
公式说明
精确率
Precision = TP / (TP + FP)
召回率
Recall = TP / (TP + FN)
F1 分数
F1 = 2 * P * R / (P + R)