DWY Time Series Decomposition Model

This paper describes a novel approach for time series decomposition suitable for time series with strong dependency on weekly cycle. There are many models of time series description. The most popular are trend + seasonal + cyclic component decomposition and the Box-Jenkins methodology. The big advantage of decomposition models is very intuitive mapping of real world into mathematical model. These models are easily understandable even to common users without deep mathematical background. On the other hand, current decomposition models are not so much suitable for modeling of complex time series with more periodic components and asymmetric behavior during the period. The model proposed in this paper better fits more deep structured time series as it respects the asymmetric behavior of the time series during all of its' periods.