Actually, there are a huge number of ways to evaluate discrete predictions with metrics, including accuracy, precision, recall, specificity, sensitivity, fallout, false omission rate, and many more. As with continuous variables, there is no one-size-fits-all metric for evaluation. Each time you approach a problem, you need to determine the metric that fits the problem and matches the goals of the project. You don't want to optimize for the wrong things and then waste a bunch of time reimplementing your model based on other metrics.
To understand these metrics and determine which is appropriate for our use case, we need to realize that there are a number of different scenarios that could ...
