← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9242 · 生成模型:GPT-6

填补统计量应从训练数据拟合

理解等级:中级

均值或中位数填补需先在训练部分估计,再用于对应验证和测试。先用整表均值填补后再划分会引入评价信息;均值若只为打印四舍五入,不应悄悄替换真正用于变换的数值。

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

      As we can see, the feature depth also has two missing values that are not numbers. It is also evident that the missing values in this feature are 2 out of 53,940 instances. This is negligible but these values can be replaced using mean imputation. This means that we will calculate the mean of the feature depth and replace all the missing values with the

    参考来源

    - 位置:《Machine Learning Fundamentals: Concepts, Models, and Applications》第05章《使用数据》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Machine Learning Fundamentals: Concepts, Models, and Applications/_epub-src/OEBPS/xhtml/17_chapter05.xhtml) - 结论依据:摘录中关于填补统计量应从训练数据拟合的定义、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:As we can see, the feature depth also has two missing values that are not numbers. It is also evident that the missing values in this feature are 2 out of 53,940 instances. This is negligible but these values can be replaced using mean imputation. This means that we will calculate the mean of the feature depth and replace all the missing values with the