State space modeling approach to decompose daily sales of a restaurant into time-dependent multi-factors

Daily sales of catering establishments are aected by several factors: The day of the week, national holidays, weather, events taking place near the site, etc. Therefore, it is beneficial to construct a model decomposing a time series of the sales into such factors. Thus, the model will allow us to predict future sales with good accuracy in order to plan various management strategies such as purchasing, sta assignments, new store openings, etc. In this study, we propose a prediction method for the daily sales of a catering establishment based on a state space model framework in which an optimal model is selected by an information criterion; we then apply it to 2-year daily sales data of a restaurant that is adjacent to a large-scale convention center and oce buildings.