flowchart TD
A[RegARIMA Modeling] --> B(Model Comparison / Diagnostics)
B --> A
A --> C[Seasonal Adjustment]
C --> D{Seasonal Adjustment <br/> Diagnostics}
D -.-> C
D -.-> A
D -.-> E[Done]
Quality assessment
This section focuses on diagnostic tools for seasonal adjustment. This will be written as a stand-alone section as well as a continuance of prior sections. The idea here is that many readers may be interested in checking the quality of their adjustments but not need help performing it.
Seasonal Adjustment Diagnostics refers to the process of evaluating and assessing the quality of seasonal adjustments made to time series data using the X-13ARIMA-SEATS software. The objectives of these diagnostics are to determine the presence of seasonality in the series, assess the overall quality of the seasonal adjustment, and make decisions regarding adjustments and options. The diagnostics include spectral graphs, stability diagnostics, sliding spans, revisions history, and monitoring and quality diagnostics. The diagnostics involve analyzing the series in the time and frequency domains, examining spectral graphs, and identifying visually significant peaks. Other diagnostic tools include the F test for seasonal regressors and the QS (autocorrelation) test for residual seasonality. Sliding spans diagnostic compares adjustments from overlapping subspans of the series to assess stability. We can use all this information to help make choice about seasonal adjustment options, such as seasonal filter lengths or sigma limits. Let’s look back at our flow chart, with a done box added, to see where seasonal adjustment diagnostics fit in. Notice there are 3 routes the diagnostics could take us; back to seasonal adjustment, to completion, or all the way back to regARIMA modeling.