There is a wide variety of performance metrics that people use in deep learning models, such as accuracy, balanced error rate, mean squared error, and many others. To keep things organized, we will divide them into three groups: for binary classification, for multiple classes, and for regression.
Binary classification
There is one essential tool used when analyzing and measuring the success of our models. It is known as a confusion matrix. A confusion matrix is not only helpful in visually displaying how a model makes predictions, but we can also retrieve other interesting information from it. The following diagram shows a template of a confusion matrix:
Figure 4.5 - A confusion matrix and the performance metrics derived from it
A confusion matrix and all the metrics derived from it are a very important way of conveying how good your models are. You should bookmark this page and come back to it whenever you need it.
In the preceding confusion matrix, you will...
