Confusion Matrix Sleuth

Inputs

Accuracy

Constrains TP + TN

Sensitivity / Recall

Constrains TP + FN

Specificity

Constrains TN + FP

PPV / Precision

Constrains TP + FP

NPV

Constrains TN + FN

Sample size (n)

Optional. Defaults to 1000.

Enter at least 3 metrics that together cover all four cells.

Quick examples

Reconstructed matrix

Sample n=— · positives —% · negatives —%

Predicted condition

Positive

Negative

Actual +

True Positive

—

False Negative

—

Actual −

False Positive

—

True Negative

—

Derived statistics

Positive class metric Negative class metric

Prevalence π

—

(TP + FN) / N

Matthews CC MCC

—

(TP·TN − FP·FN) /
√((TP+FP)(TP+FN)(TN+FP)(TN+FN))

F1 score F1

—

2·PPV·TPR / (PPV + TPR)

01 Predictive values

Precision PPV

—

TP / (TP + FP)

Neg. Pred. Value NPV

—

TN / (TN + FN)

False Disc. Rate FDR

—

FP / (FP + TP) = 1 − PPV

False Omission FOR

—

FN / (FN + TN) = 1 − NPV

02 True / False rates

Sensitivity TPR

—

TP / (TP + FN)

Specificity TNR

—

TN / (TN + FP)

False Pos. Rate FPR

—

FP / (TN + FP) = 1 − TNR

False Neg. Rate FNR

—

FN / (TP + FN) = 1 − TPR

03 Likelihood & combined

Pos. Likelihood LR+

—

TPR / (1 − TNR)

Neg. Likelihood LR−

—

(1 − TPR) / TNR

Diagnostic Odds DOR

—

LR+ / LR−

Youden's J J

—

TPR + TNR − 1

04 Aggregate scores

Accuracy ACC

—

(TP + TN) / N

Balanced Accuracy BACC

—

(TPR + TNR) / 2

Error Rate ERR

—

(FP + FN) / N = 1 − ACC

Negative Prev. 1−π

—

(TN + FP) / N