Unit U4.3 · embed after: the 16-image worked example

Confusion Matrix - six metrics, five presets, sliders and error costs

Data Science & Big Data Analytics • ITec4121 • Wolkite University • Online-First

Fill TP/FP/TN/FN with number boxes or 0–100 sliders, or load a preset - the 16-image exam example, email-spam, COVID screening, loan default and a low-prevalence cancer screen. Get Accuracy, Precision, Recall, Specificity, NPV, Youden’s J and F1, plus error rate, prevalence and a weighted error cost with your own FP/FN costs.

Matrix & metrics

Preset:

Enter counts:

TP FP TN FN
TP FP TN FN Sliders cover 0–100; use the boxes for the 500/1000/2000-row presets.
FP cost FN cost

	Actual +	Actual −
Predicted +	TP=9	FP=1
Predicted −	FN=3	TN=3

Formula sheet (memorise)

Metric	Formula	16-image values
Accuracy	(TP+TN)/(TP+FP+TN+FN)	12/16 = 0.75
Precision	TP/(TP+FP)	9/10 = 0.90
Recall	TP/(TP+FN)	9/12 = 0.75
F1	2·P·R/(P+R)	0.82