Build a labeled binary confusion-matrix report with precision, recall, specificity, F1, balanced accuracy and Matthews correlation coefficient.
How to use this tool
- Choose the event or model and review the assumptions below.
- Enter the labeled inputs.
- Select Calculate and read the labeled outputs.
- Change an input and calculate again; Reset restores the example inputs.
Calculation method
Metrics are calculated from a labeled actual-by-predicted binary table. MCC = (TP·TN−FP·FN) / √[(TP+FP)(TP+FN)(TN+FP)(TN+FN)].
Assumptions and limits
Binary nonnegative integer counts. Undefined ratios stay undefined; no hidden replacement with zero. This report does not select a threshold, certify model fairness or estimate uncertainty.
Worked example
For TP = 80, FP = 10, FN = 20 and TN = 90, actual positives total 100 and predicted positives total 90. Precision is 80/90, approximately 88.8889%, while recall is 80/100 = 80%. F1 is 160/190, approximately 84.2105%. The different denominators explain why the rates differ. Metrics with zero denominators stay undefined rather than silently becoming zero.
Worked reading and practical questions
- Label the Positive Class and Table Orientation Before Reporting Metrics
- MCC, F1 and Specificity: Report the Tradeoffs Hidden by One Score
Browse Statistics calculators or explore other tools on the GreenCalc homepage.
Frequently asked questions
What does this calculator assume?
Binary nonnegative integer counts. Undefined ratios stay undefined; no hidden replacement with zero. This report does not select a threshold, certify model fairness or estimate uncertainty.
How do I enter and update the inputs?
Use the labeled fields, then select Calculate. Percent fields use 0 to 100. After editing, calculate again; Reset restores the example inputs.