A confusion matrix tabulates predicted classes against true classes. Each cell counts one outcome, making classifier errors visible instead of hiding them behind accuracy—especially when class frequencies are uneven in ligand screening. section 3.3.3
For binary classification, the four cells represent:
- True positive (TP): a positive instance predicted correctly. section 3.3.3
- False negative (FN): a positive instance predicted as negative. section 3.3.3
- False positive (FP): a negative instance predicted as positive. section 3.3.3
- True negative (TN): a negative instance predicted correctly. section 3.3.3
Reading a classifier’s errors
The matrix [[119, 1], [13, 259]] records 119 true negatives, 1 false negative, 13 false positives, and 259 true positives. These counts support precision and recall, which reveal error patterns that accuracy can obscure. section 3.3.3















