While the scope of the book does not allow for a complete treatment of the confusion matrix, a short explanation and a link are provided as a quick reference.
The confusion matrix is just a fancy name for an error matrix. It is mostly used in unsupervised learning to visualize the performance. It is a layout that captures actual versus predicted outcomes with an identical set of labels in two dimensions:
Confusion Matrix
To get a quick introduction to the confusion matrix in unsupervised and supervised statistical learning systems, see https://en.wikipedia.org/wiki/Confusion_matrix.