According to Japan’s Ministry of Agriculture, Forestry and Fisheries, 5.7 million tons of food was discarded in 2019. With Japan’s population of 126 million, that roughly translates to the equivalent of one bowl of rice per person, per day.
One of the major contributors to this monumental amount of food waste is an unbalanced supply and demand equation between restaurants, stores and food suppliers. Analog-style operation methods have in part led to unsold inventories and rising costs associated with food disposal.
To address this serious social issue, and to help retail operators better allocate staff, SoftBank Corp. (TOKYO: 9434) launched a new service in collaboration with the Japan Weather Association. The service, called "Sakimiru," which means “see the future,” is an AI-powered demand forecasting service that utilizes population flow data and weather data. To provide highly accurate demand forecasts, SoftBank and the Japan Weather Association jointly developed an AI algorithm that utilizes multiple data sources.
Sakimiru data source examples
- Statistical, anonymized population flow data taken from SoftBank’s base stations, which have access to historical mobile phone location data
- The Japan Weather Association’s weather data
- Sales and customer number data for stores operated by clients
First service offering provides customer traffic forecasts
In its first phase, Sakimiru will provide functions that predict customer traffic at the individual stores of clients. With more accurate customer forecasts, clients will be able to optimize their product orders and work shifts, which will in turn lead to reduced food waste and better personnel allocations.
With Sakimiru, clients can access customer traffic forecasts based on weather data for each of their stores up to two weeks in advance. Weather forecasts are normally available for wide areas such as cities, towns or villages, but with weather data from the Japan Weather Association, Sakimiru clients can access forecasts for individual store locations.
Sakimiru also uses statistical information on floating populations, or the flow of people. This data comes from SoftBank Corp.’s base stations, which have access to tens of millions of mobile phone locations, which is then extrapolated to Japan’s population of 126 million for predictive purposes. Using this statistical, anonymized data, clients can access customer traffic predictions without relying solely on the past histories of their own stores.
Taking these various data sources together, Sakimiru clients will be able to place orders and draw up work shift schedules with greater precision.
Prior to the launch of Sakimiru, trials were conducted at 10 "Vdrug" stores, a drug store chain operated by Chubu Yakuhin Co., Ltd., in cooperation with owner Valor Holdings Co., Ltd. At participating stores, Sakimiru was found to be highly effective, with an average customer traffic prediction accuracy of 93%.
Looking ahead, SoftBank and the Japan Weather Association will continue to develop new demand forecasting functions for placing orders and creating work shift schedules. Through Sakimiru, both will aim to contribute to the achievement of the SDGs by reducing food waste and enhancing productivity.
(Posted on February 15, 2022)
by SoftBank News Editors