← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9246.c · 生成模型:GPT-6

VIF衡量被其余列解释程度

理解等级:中级

VIF衡量某预测列被其他列解释时带来的系数方差膨胀,常以1/(1−Rⱼ²)表达。阈值5或10只是经验提醒,需结合目标与样本;两两相关低也可能存在多列线性依赖。

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

      Detecting Multicollinearity Using Variance Inflation Factor (VIF): MINITAB provides an option to calculate VIF for each predictor variable that measures how much the variance of the estimated regression coefficients are inflated as compared to when the predictor variables are not linearly related. Use the guidelines in Table 6.6 to interpret the VIF.

    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VIF衡量被其余列解释程度的定义、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:Detecting Multicollinearity Using Variance Inflation Factor (VIF): MINITAB provides an option to calculate VIF for each predictor variable that measures how much the variance of the estimated regression coefficients are inflated as compared to when the predictor variables are not linearly related. Use the guidelines in Table 6.6 to interpret the VIF.