Case Study

Estimating a Benchmark for Ticket Routing

Scenario: A product team is building a model that assigns incoming support tickets to one of several specialist queues. As part of project planning, the team tries to use experienced agents as a reference point, but the agents often disagree about which queue a ticket should go to.

Question: What does this suggest about estimating the best achievable error rate for the system?

Sample answer: It suggests that estimating the best possible error rate will be difficult, because people do not reliably agree on the correct label for this task.

Key points:

  • The task is ambiguous enough that trained people disagree.
  • Without a dependable human reference, the ceiling on performance is hard to pin down.
  • Therefore, the optimal error rate is difficult to estimate.

Rubric: The response must state that strong disagreement among human judges makes the best achievable error rate hard to estimate.

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