知识卡片
极端值需要区分真实尾部与错误
少见或距离均值远的记录可能是真实对象,也可能是录入或测量错误。先核实来源和任务价值再决定修正、保留或排除;少于1%本身不能证明删除不会扭曲目标人群。
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
- · 使用数据
原书出处 · 1 处
- · 查看原书依据
A data set that has missing values, or has outliers, is considered as messy data set. The outliers are usually very small or large value in a data set that are unusual information. It is important to deal with these issues using data preprocessing or transformation. It is not unusual to find all missing values from one or more instances or where some instances or
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:摘录中关于极端值需要区分真实尾部与错误的定义、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:A data set that has missing values, or has outliers, is considered as messy data set. The outliers are usually very small or large value in a data set that are unusual information. It is important to deal with these issues using data preprocessing or transformation. It is not unusual to find all missing values from one or more instances or where some instances or