← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9250.b · 生成模型:GPT-6

树对异常值并非完全免疫

理解等级:中级

特征阈值树可能对部分单调变换或极端特征幅值较不敏感,但异常标签仍可改变分裂和叶预测,回归均值也可能受极端结果影响。不能从不依赖欧氏距离推断异常值毫无意义。

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

      They are not affected by outliers. Outliers have no meaningful effect on CART. Limitations of CART May result in overfitting. May result in high variance and low bias. Applications of CART Classification algorithms are useful models for categorical data. Has applications in finance, blood donor classification, and others. Difference Between Classification and Regression Tree A classification tree is an algorithm where the target variable is

    参考来源

    - 位置:《Machine Learning Fundamentals: Concepts, Models, and Applications》第06章《监督学习模型》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Machine Learning Fundamentals: Concepts, Models, and Applications/_epub-src/OEBPS/xhtml/19_chapter06.xhtml) - 结论依据:摘录中关于树对异常值并非完全免疫的定义、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:They are not affected by outliers. Outliers have no meaningful effect on CART. Limitations of CART May result in overfitting. May result in high variance and low bias. Applications of CART Classification algorithms are useful models for categorical data. Has applications in finance, blood donor classification, and others. Difference Between Classification and Regression Tree A classification tree is an algorithm where the target variable is