Selecting a Dev Set Size for a Small Accuracy Gain
Case context: A product team is improving a fraud detection classifier and expects future model changes to raise validation accuracy by about 0.1%. Their development set contains 1,000 labeled examples.
Question: Assess whether a dev set with 1,000 examples is large enough to notice a 0.1% improvement, and state a concrete dev set size that would make such a difference more detectable.
Sample answer: A dev set of 1,000 examples is not large enough to dependably notice a 0.1% improvement. With only 1,000 cases, a 0.1% shift corresponds to a single example, which is too little to separate from ordinary variation. A dev set of about 10,000 examples would give a much better chance of detecting a 0.1% gain.
Key points:
- A 1,000-example dev set is still too small for a 0.1% improvement, even though it may be a reasonable general size.
- On a set of 1,000 examples, a 0.1% change is equivalent to just 1 example.
- Expanding the dev set to 10,000 examples is a practical target for noticing a 0.1% improvement.
Rubric: The student must explain that 1,000 examples is too small to reliably observe a 0.1% improvement and recommend increasing the dev set to 10,000 examples to make that change easier to detect.
0
1
Tags
Machine Learning
Deep Learning
Machine Learning Strategy
Supervised Learning
Dive into Deep Learning @ D2L
Data Science
Machine Learning Yearning @ DeepLearning.AI
Related
What dev set size is commonly suggested for detecting a 0.1% improvement in accuracy?
True or False: A validation set with 1,000 examples is enough to reliably detect a 0.1% improvement in accuracy.
Development sets with _____ to 10,000 examples are often considered common.
Which dev set size range is commonly used in practical machine learning projects?
True or False: A dev set with 10,000 examples can often reveal an accuracy gain of about 0.1 percentage point.
Validation Set Size for Tiny Accuracy Gains
Match each development-set idea to its best description.
Put the validation-set sizing logic in order.
What size of accuracy change is a 10,000-example dev set generally able to detect?
True or False: A development set with 1,000 examples is smaller than the commonly suggested range for spotting very small improvements.
Common development-set sizes begin at _____ examples.
Match each development-set situation to the most accurate classification.
Arrange the steps for deciding when a dev set must be large enough to notice a tiny metric gain.
Why Dev Set Size Matters for Detecting Small Gains
Selecting a Dev Set Size for a Small Accuracy Gain
Choosing a Development Set for Tiny Metric Gains