← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9262.d · 生成模型:GPT-6

排序AUC不直接评价部署阈值

理解等级:中级

ROC AUC按正负样本的分数排序衡量区分能力,需连续得分或概率,而非只输入固定阈值的类别。常数得分为0.5不表示所有0.5模型都相同;类别失衡时还应看精确率召回及实际阈值。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      Area Under ROC Curve (AUC) (for binary classification): the AUC quantifies the ability of a binary classifier to distinguish between two classes. It is desirable when the data set is imbalanced (i.e., the data set has significantly more samples corresponding to one class than the other). The AUC ranges between 0 and 1. A value close to 1 indicates that the model is able to

    参考来源

    - 位置:《Machine Learning Fundamentals: Concepts, Models, and Applications》第09章《训练机器学习模型》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Machine Learning Fundamentals: Concepts, Models, and Applications/_epub-src/OEBPS/xhtml/23_chapter09.xhtml) - 结论依据:摘录中关于排序AUC不直接评价部署阈值的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:Area Under ROC Curve (AUC) (for binary classification): the AUC quantifies the ability of a binary classifier to distinguish between two classes. It is desirable when the data set is imbalanced (i.e., the data set has significantly more samples corresponding to one class than the other). The AUC ranges between 0 and 1. A value close to 1 indicates that the model is able to