Confusion Matrix

Confusion Matrix

We are now ready to lay down a first numerical overview of our classification performance, also called contingency table.

<math>Y</math> <math>P</math> <math>N</math> <math>\widehat{Y}</math> <math>\widehat{P}</math> TP FP <math>\widehat{N}</math> FN TN

This will give us a broad overview of how well our classifier is doing. In general we want the 'right guesses' T to be big and the 'mistakes' to be small.

Confusion matrices can also be created for n-classes classifiers:

The main diagonal will then be the 'right guesses' and the rest will be the different type of 'mistakes'.

Confusion Matrix - Wikipedia