Confusion matrix examples.
Same tool, four requests. Every drawing below is a real render.
Rather start from a finished drawing? Browse 17 heatmap templates →
What a confusion matrix is.
A confusion matrix is a table for evaluating a classification model. Rows represent the actual class and columns represent the predicted class. Each cell counts how often the model made that actual-versus-predicted outcome.
The diagonal contains correct classifications. Off-diagonal cells reveal the specific mistakes, such as billing tickets sent to technical support. From these counts you can calculate accuracy, precision, recall, specificity and F1 score.
- Engine
- echarts-heatmap
- Editable
- Describe the change
- Export
- SVG · PNG · PDF
Who uses confusion matrices.
Inspect which classes a routing model confuses before changing features or thresholds.
Discuss false approvals and false alerts in a fraud-screening model.
Connect classification counts to accuracy, precision and recall in a visual form.
How to make a confusion matrix in three steps.
Describe it
One paragraph is enough to start.
See the drawing
Drawn by the right engine.
Say what changes
Every edit keeps a version.
Common questions
What is a confusion matrix?
A confusion matrix is a performance table for a classification model. It counts actual classes against predicted classes, separating correct predictions from each kind of error.
How do I read a confusion matrix?
Read the diagonal as correct predictions and the off-diagonal cells as errors. Large off-diagonal values identify the classes the model confuses most often.
Can a confusion matrix have more than two classes?
Yes. An N-class model uses an N by N matrix, with one row and one column for each class.
Which metrics can I calculate from a confusion matrix?
Common metrics include accuracy, precision, recall, F1 score and specificity. The appropriate metric depends on which error type matters most.
Other drawings for the same work.
Make a confusion matrix now.
Free account, no card. Provide the actual and predicted class counts.
Open the editor