Each model, no matter how complex and accurate, makes mistakes. It is natural to expect that some models will be better than others when solving a specific problem. Currently, we can measure errors by comparing individual model predictions with the ground truth. It would be useful to summarize them into a single number for measuring the model's performance. We can use a metric to do this. There are many kinds of metrics that are suitable for different machine learning problems.
In particular, for regression problems the most common metric is the root mean square error, or RMSE:
Let's examine the elements of this formula: ...
