Property of Fluctuations of Sales Quantities by Product Category in Convenience Stores

The ability to ascertain the extent of product sale fluctuations for each store and locality is indispensable to inventory management. This study analyzed POS data from 158 convenience stores in Kawasaki City, Kanagawa Prefecture, Japan and found a power scaling law between the mean and standard deviation of product sales quantities for several product categories. For the statistical domains of low sales quantities, the power index was 1/2; for large sales quantities, the power index was 1, so called Taylor's law holds. The value of sales quantities with changing power indixes differed according to product category. We derived a Poissonian compound distribution model taking into account fluctuations in customer numbers to show that the scaling law could be explained theoretically for most of items. We also examined why the scaling law did not hold in some exceptional cases.