← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8779 · 生成模型:GPT-6

通用分箱不自动保留原边界

理解等级:高级

仅传箱数不能保证通用分箱与原手写年龄或票价阈值一致,等宽、等频和固定边界会产生不同特征。若当前是保持行为的重构,应保存边界、端点和缺失处理;有意换方案需另做质量验证。

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

      implementation detail of converting Age from a continuous variable (0 to 100) to a binned ordinal variable (0, 1, 2, 3, and 4), we later realized that this function could be easily extended (without modification) for a similar data transformation on the Fare column. categorize_column() was extensible—we can specify how many bins we want using the num_bins argument—without the need to modify the function’s implementation:

    参考来源

    - 位置:《Effective Machine Learning Teams (David Tan, Ada Leung, David Colls)》第08章《重构与技术债》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Effective Machine Learning Teams (David Tan, Ada Leung, David Colls)/_epub-src/OEBPS/ch08.html) - 结论依据:摘录中关于通用分箱不自动保留原边界的定义、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:implementation detail of converting Age from a continuous variable (0 to 100) to a binned ordinal variable (0, 1, 2, 3, and 4), we later realized that this function could be easily extended (without modification) for a similar data transformation on the Fare column. categorize_column() was extensible—we can specify how many bins we want using the num_bins argument—without the need to modify the function’s implementation: