In simple words: A confusion matrix shows the kinds of errors, not just one overall score.
The four outcomes
True positive: predicted yes and actually yes. True negative: predicted no and actually no. False positive: predicted yes but actually no. False negative: predicted no but actually yes. Decide what “positive” means before reading the table.
Why it helps
These four counts show which mistakes the system makes. From them we can calculate precision, recall, and accuracy. In a health screening task, a false negative can be particularly serious, so the team may give recall more attention.
See the flow
If a model misses 3 of 20 real fraud cases, those 3 are false negatives. If it wrongly blocks 5 genuine payments, those 5 are false positives.
Remember: A confusion matrix shows the kinds of errors, not just one overall score.