The request
“Create a confusion matrix heatmap for helpdesk routing with actual rows and predicted columns Billing, Technical and Account. Use counts: Billing 124, 18, 6; Technical 14, 109, 21; Account 5, 17, 96.”
Then tryChange the Technical predicted as Account count to 12.Add a subtitle for the April validation set.
What is in this drawing
Read the decisions behind it.
01Name both axes
Separate actual from predicted labels.
02Show every count
Reveal each error type.
03Inspect off-diagonal cells
Prioritize routing mistakes.
Use for a classification model with stable, labeled validation data.
More like this
Other examples in this family.
FAQ
Common questions
What does the diagonal show?
It shows tickets routed to their actual class correctly.
Which cells are errors?
Every off-diagonal cell is a misrouted ticket count.
Why use a validation set?
It tests performance on labeled data not used to train the model.