知识卡片
划分比例需同时考虑精度与覆盖
60/20/20和80/10/10是示例,比例应结合总样本、少数类别、群组和时间结构确定。更小评价集常使估计不确定性增大,不等于必然更有偏;增加训练数据也不保证准确率单调提高。
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
- · 查看原书依据
Dividing the data set accurately is critical to creating a successful and effective ML model. Dividing the data set in training, validation, and test sets is about determining the size of each set. While more training data should lead to a more accurate model, a smaller amount of test and validation data may provide a more biased estimate of the model and its performance. So,
参考来源
- 位置:《Machine Learning Fundamentals: Concepts, Models, and Applications》第03章《机器学习流程与数据管理》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Machine Learning Fundamentals: Concepts, Models, and Applications/_epub-src/OEBPS/xhtml/14_chapter03.xhtml)
- 结论依据:摘录中关于划分比例需同时考虑精度与覆盖的定义、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:Dividing the data set accurately is critical to creating a successful and effective ML model. Dividing the data set in training, validation, and test sets is about determining the size of each set. While more training data should lead to a more accurate model, a smaller amount of test and validation data may provide a more biased estimate of the model and its performance. So,