Case Study

How a reviewer should log a set of misclassified support tickets.

Case context: A reviewer is checking roughly 100 support tickets that a model assigned to the wrong class. The goal is to spot repeated error patterns without losing memorable details from individual tickets.

Question: What is a good way to record the review, and which fields should the record include?

Sample answer: Use a spreadsheet and update it during the review. Include columns such as Ticket ID, Billing issue, Shipping issue, Hard to read, Wrongly tagged, and Notes. The issue columns make it easy to compare patterns across tickets, and the Notes column preserves unusual details from specific cases.

Key points:

  • Use a spreadsheet as the review log.
  • Complete it while checking about 100 wrong predictions.
  • Add columns for Ticket ID, Billing issue, Shipping issue, Hard to read, Wrongly tagged, and Notes.
  • Use Notes for details that help you remember specific tickets.

Rubric: The response should recommend a spreadsheet used during the review, name the supported fields, and explain the purpose of notes. Do not give full credit if the answer leaves out either the structured columns or the case-specific notes.

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