Understanding evaluation metrics

In machine learning, an evaluation metric is a measure used to quantify the quality of a model’s predictions.

If understood and interpreted correctly, they can provide you with a measure with which to evaluate the quality of a model and, therefore, make more informed decisions about its use or whether more work is needed to train a more accurate model.

There is a wide range of evaluation metrics within machine learning, and different types of machine learning models require different evaluation metrics.

When considering supervised machine learning, which we covered in Chapter 7, there are two groups of models: regression models and classification models, each with its own set of evaluation metrics.

First, let’s look at some of the more common metrics used for evaluating regression models.

Evaluating regression models

Imagine you’re a retail executive trying to forecast the next quarter’s sales. You’...