- Print
- DarkLight
- PDF
A table that visualizes the performance of a classification algorithm, showing how often it correctly or incorrectly predicts different categories.
Rows: Represent the "True" or actual categories of the data.
Columns: Represent the "Predicted" categories assigned by the skill.
Numbers within the cells: Indicate the count of how many instances fall into each combination of True and Predicted categories.
Diagonal Cells (from top right to bottom left):
These represent the cases where the True category matches the Predicted category - these are the correct predictions.
The higher the numbers in these cells, the better your model is at classifying those specific categories.
Off-Diagonal Cells:
These represent the cases where the True category and Predicted category don't match - these are the errors or misclassifications.
Analyzing these cells helps you understand which categories your model tends to confuse with each other.