知识卡片
分类指标应匹配错误代价
准确率只计算正确数量比例,无法独自说明少数类或不同错误成本。log loss评价真实标签概率,F1综合precision与recall,AUC需指明曲线;它们不分别等同假阳性计数或精密率。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
- · 机器学习基础
原书出处 · 1 处
- · 查看原书依据
Classification accuracy (which is the ratio of correct prediction divided by total number of input samples). The percentage gives the model accuracy. Logarithmic loss or log loss. It works by penalizing the false (false positive) classification. Area under curve: it is one of the widely used metrics used for binary classification. F1 score: F1-scores are used to evaluate the performance of the classification algorithms. This
参考来源
- 位置:《Machine Learning Fundamentals: Concepts, Models, and Applications》第02章《机器学习基础》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Machine Learning Fundamentals: Concepts, Models, and Applications/_epub-src/OEBPS/xhtml/13_chapter02.xhtml)
- 结论依据:摘录中关于分类指标应匹配错误代价的定义、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:Classification accuracy (which is the ratio of correct prediction divided by total number of input samples). The percentage gives the model accuracy. Logarithmic loss or log loss. It works by penalizing the false (false positive) classification. Area under curve: it is one of the widely used metrics used for binary classification. F1 score: F1-scores are used to evaluate the performance of the classification algorithms. This