← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6481.h.13 · 生成模型:GPT-6

重要性分数不能随意聚合

理解等级:高级

不同模型的重要性可能有不同尺度与含义。为求几何平均而把负分数随意转成正值,并不能建立共同量纲;1减p值也不是可与其他重要性直接比较的效应大小。本章这种非标准聚合不足以证明某属性最重要。

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

      negative values in importance are disallowed in the geometric mean. So, in order to cancel the negativity, each negative value is transformed into a positive number by dividing by 10 and multiplying by -1, which makes a negative number lower than the least positive number. Further, the features are sorted in reverse order with respect to geometric mean and also categorized to have a sense

    参考来源

    - 位置:《Artificial Intelligence and Data Science in Recommendation System: Current Trends, Technologies, and Applications》第12章《时尚零售评估与预测》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Artificial Intelligence and Data Science in Recommendation System: Current Trends, Technologies, and Applications/_epub-src/OEBPS/Text/9789815136746-C12.html) - 结论依据:摘录中关于重要性分数不能随意聚合的定义、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:negative values in importance are disallowed in the geometric mean. So, in order to cancel the negativity, each negative value is transformed into a positive number by dividing by 10 and multiplying by -1, which makes a negative number lower than the least positive number. Further, the features are sorted in reverse order with respect to geometric mean and also categorized to have a sense