Evaluating models

In this section, we will look at metrics for evaluating how well a model is performing. This section focuses on metrics to use to evaluate how well a model predicts a target variable in binary classification. We will discuss how to compute accuracy, precision, recall, the F1 score, and the Bayes factor, along with how to interpret each of these metrics.

Accuracy

Accuracy measures how frequently an algorithm predicted the correct label. On the surface, this looks like a good enough metric, but accuracy alone does not convey the quality of an algorithm. A problem could have an algorithm that is very accurate, but only because the learning problem is, in some sense, easy, such as predicting on any particular...