Abstract
The supermarket industry is constantly examining how to reduce waste and improve net profit in difficult circumstances. To date, many prediction methods have been developed for reviewing the allocation of employees and part-timers by predicting the number of store visitors. However, the number of store visitors generally also includes accompanying visitors, and does not only consist of people making payments at cash registers. The present study proposes a system that, based on live POS data provided by an actual supermarket chain and examining rental cash registers, predicts the optimal number of cash registers in operation for each time band several weeks after introducing register operation rates and numbers of visitors leaving per unit time. The present paper proposed a method to obtain data including the optimal number of cash registers in operation from the POS data. It also investigated the factors in fluctuating customer numbers in concrete terms. Moreover, it evidenced that the optimal number of cash registers in operation hardly changes after 1-2 weeks. This allows for the appropriate planning of cash register staff numbers per time unit. Furthermore, it enables stores to obtain the optimal number of cash registers. The store whose POS data were used had 7 cash registers, but it was ascertained that when the optimal operation rates were 80 %, 70 %, and 60 %, sufficient cash register numbers were 4, 5, and 6, respectively. Opening hours for this store were between the 7 am and 10 am time bands. Virtually no customers visited the store in the 7 am, 8 am, and 10 pm time bands regardless of the day of the week, and it was concluded that the opening hours should be changed to the 9 am - 9 pm time bands.
