Multiclass confusion matrix reader
Identify the weakest class recall and compare macro recall with overall accuracy.
Your entries stay in this page’s memory and are not submitted to us. Downloads are generated locally. Reloading or leaving the page clears your entries. Do not enter patient information.
Enter 2–30 comma-separated counts per row to form a square matrix. Rows are actual classes and columns are predicted classes, in the same class order. Class 1 means the first row and column. Every actual class needs at least one observation; diagonal cells are correct predictions.
The starting values are hypothetical. Results update as you edit. Use aggregate or synthetic inputs only.
Calculation results
- Overall accuracy
- 0.75
- Macro recall
- 0.68333333
- Recall by class
- 1: 0.8, 2: 0.75, 3: 0.5
How to interpret the output
Use the same class order on both axes. Every actual class needs observations. Macro recall weights each class equally while accuracy weights each observation equally. Counts must describe one disjoint evaluation set.
Method reference