We need to use a different set of metrics for evaluating regression models from those for classification model evaluations. This is because the prediction output of a regression model takes continuous values, meaning it can take any value and is not restricted to taking from a predefined set of values. On the other hand, as we have seen in Chapter 8, Predicting the Likelihood of Marketing Engagement, the prediction output of a classification model can only take a certain number of values. As was the case for the engagement prediction, our classification model from the previous chapter could only take two values—zero for no engagement and one for engagement. Because of this difference, we need to use different metrics to evaluate regression models.
In this section, we are going to discuss four commonly used methodologies to evaluate regression...
