Short Answer

Why a 2,000-Item Manual Review Set Fits a 5% Error Rate

Question: In one or two sentences, explain why a manual review development set should be about 2,000 examples large when a classifier’s error rate is 5%.

Sample answer: With a 5% error rate, a set of 2,000 examples will contain about 100 mistakes, since 0.05 × 2,000 = 100. That gives enough failed cases to inspect patterns instead of trying to draw conclusions from only a handful of errors.

Key points:

  • 0.05 × 2,000 = 100
  • The purpose is to collect roughly 100 errors for analysis

Rubric: The response should state the calculation for 5% of 2,000 or explain that the set is sized to yield about 100 errors for inspection.

0

1

Updated 2026-08-12

Contributors are:

Who are from:

Tags

Machine Learning

Deep Learning

Machine Learning Strategy

Supervised Learning

Dive into Deep Learning @ D2L

Data Science

Machine Learning Yearning @ DeepLearning.AI