The decomposition of a given time series into trend, seasonal component, and irregular component is formulated as a minimization problem. The trend is chosen such that it is as smooth as possible; the seasonal component is chosen such that it exhibits a seasonal pattern as stable as possible; and...

The paper discusses a new seasonality hypothesis which is one part of a weighted regression approach for the decomposition of a time series into a trend, a seasonal component and an irregular component. It is shown that there exists a regression formulation leading, as in the descriptive...