Essay

Explain the iterative nature of error analysis.

Question: Based on Ng's explanation in Machine Learning Yearning, explain why error analysis is considered an iterative process. Detail the typical cycle a practitioner goes through when defining error categories for misclassified examples.

Sample answer: Error analysis is iterative because categories are not strictly predefined. A practitioner often starts with no categories, examines a few examples to brainstorm initial categories, and begins manual categorization. During this process, new categories often emerge, requiring the practitioner to revisit and re-examine previously analyzed examples under these new categories.

Key points:

  • Categories do not need to be predefined.
  • Initial examples inspire the first set of categories.
  • Manual categorization often reveals the need for new categories.
  • Previous examples must be re-examined when new categories are added.

Rubric: Full credit for mentioning starting without categories, brainstorming after seeing initial examples, discovering new categories during manual categorization, and re-examining examples with new categories.

0

1

Updated 2026-05-27

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