Residuals are differences between observed data points and the values predicted by the model. They represent the "leftover" variation in the data that the model did not capture.

Analyse these residuals to assess the quality of a time series model and diagnose potential issues.