When evaluating a model, we have numerical metrics and visualization techniques that are complementary ways to asses model performance. In this section, we will go back to our diamond prices problem and we will talk about the most common metrics and plots that are used to evaluate regression models. We will also define our own evaluation metric in the context of the business problem we are trying to solve.
Before we begin, I would like to make a clarification—in this section, we use interchangeably the terms errors and residuals in reference to the difference between actual and predicted value: actual_price – predicted_price. Technically, the term error refers to a population concept and has to do with a theoretical population value. So, although technically we should not use the term errors meaning residuals, for the sake of clarity...
