StructuralDecompose is a method that breaks a time series algorithm into various parts. It is particularly well suited to a time series that has several level shifts within it.

StructuralDecompose returns the series constituent parts including its Trend, Seasonality and residuals. As well as a fairly inbuilt summary of the time series itself and how well it has fit the data. As it performs inbuilt Anomaly Detection, it also returns a series of points that it considers to be anomalies. However more advanced anomaly detection techniques should be considered if you are doing anomaly detection on the time series.