知识卡片
为每个新特征计算投资回报率(ROI)
新特征是否该保留,不能只看它是否提升模型质量。作者提醒:即便一个特征确实带来价值,也可能不值得为采集、处理、存储和维护它付出代价;因此除极少数显而易见的情形外,应为每个新增特征粗略计算一次投资回报率(ROI),以避开『有用但成本高于收益』的特征。评估可以很轻量:在已有模型上只加一个特征重训,再把少量流量(如 1% 的请求)导向新模型做 A/B 测试,看业务指标是否改善——例如为推荐模型加入『用户所在国家』后,被推荐商品是否更常被真正购买。第 5 章会更系统展开,本章要点是让收益与成本一起进入决策。
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
原书出处 · 1 处
- Feature and Training Data · 查看原书依据
Keep in mind that even a feature that adds value may not be worth the cost to collect, process, store, and maintain it. For all but the most trivially obvious features, it is a good habit to calculate a rough return on investment (ROI) for every new feature added to the system. This will help us avoid useful features that are still more expensive than the value that they add.
参考来源
- 位置:《Reliable Machine Learning》第04章《Feature and Training Data》(源文件:extracted/Reliable Machine Learning/_epub-src/OEBPS/ch04.xhtml)
- 结论依据:原文明确阐述该主题,并给出定义、实践或反例。
- 原始内容:Keep in mind that even a feature that adds value may not be worth the cost to collect, process, store, and maintain it. For all but the most trivially obvious features, it is a good habit to calculate a rough return on investment (ROI) for every new feature added to the system. This will help us avoid useful features that are still more expensive than the value that they add.