Diagnostic Checks

Like many statistical models, ARIMA models are based on assumptions about the data they are trained on. For instance, the error component is assumed to be Gaussian white noise. After training an ARIMA model, we need to confirm that these assumptions hold. To do that, we perform diagnostic checks.

Diagnostic checks play a crucial role in ARIMA modeling as they help validate the model assumptions and identify any shortcomings. There are several diagnostic checks available, each serving a unique purpose and shedding light on different aspects of the model’s performance. These checks allow us to assess the model’s fit, identify any remaining patterns or systematic errors in residuals, and evaluate the significance of model parameters. Ultimately, by performing diagnostic checks, we ensure that our model is reliable and provides accurate forecasts.

Model summary

The ARIMAResult object in statsmodels comes with a very handy method to explore the fitted model. It is called summary(), and it does exactly what it says on the tin. It prints a summary of the model. Let’s see how it works with the models that we’ve trained in previous lessons.