Case Study

Diagnosing a Training–Validation Gap

Case context: You are building a spam classifier. It performs well on the training emails, but its accuracy drops sharply on a held-out validation set, suggesting that the training data and validation data may not match well.

Question: According to basic error-analysis practice for this kind of mismatch, what should you compare to identify the underlying cause?

Sample answer: You should compare the training set and the validation set so you can identify how they differ and determine whether those differences explain the performance drop.

Key points:

  • Compare the training set with the validation set.
  • Look for meaningful differences between the two sets.
  • Use those differences to explain the mismatch.

Rubric: The answer must say that the analysis should examine differences between the training set and the validation set.

0

1

Updated 2026-08-12

Contributors are:

Who are from:

Tags

Machine Learning

Deep Learning

Supervised Learning

Dive into Deep Learning @ D2L

Data Science

Machine Learning Strategy

Machine Learning Yearning @ DeepLearning.AI