Launched "SAKIMIRU", an AI demand forecasting service for the retail and restaurant industries that utilizes people flow and weather data, etc. -Provides highly accurate customer traffic forecasting function at a low price and contributes to optimizing the number of orders and work shifts- Press Release2022.01.31

January 31, 2022 SoftBank Corp. Japan Weather Association

SoftBank Corp. (hereinafter "SoftBank") and the Japan Weather Association (hereinafter "Japan Weather Association") will provide AI (artificial intelligence) that utilizes people flow and weather data for the retail and restaurant industries. The demand forecast service “Sakimir” will be jointly developed, and SoftBank will start providing it from January 31, 2022. At the beginning of the service, we will provide a function of "predicting the number of customers visiting the store" * 1 that predicts the number of customers visiting each store. As a result, companies can adjust the number of product orders and employee work shifts for each store based on the number of customers predicted by Sakimil. "Sakimir" will be introduced sequentially from January 31, 2022, targeting a total of about 1,200 stores operated by group companies of Valor Holdings Co., Ltd., which operates supermarkets and drugstores mainly in the Chubu region. I plan to In addition, Godiva Japan Co., Ltd. is considering the introduction of "Sakimill" for about 300 stores operated by the company.

"Sakimir" is based on the location information data of the terminal obtained from SoftBank's mobile phone base station * 2, the weather data held by the Japan Weather Association, and the store owned by the introducing company. This is a service*3 that analyzes various data such as sales and the number of customers visiting stores using an AI algorithm jointly developed by SoftBank and the Japan Weather Association to make highly accurate demand forecasts. First of all, we will provide the function of "predicting the number of customers visiting the store" at a low price of 5,390 yen (including tax) per store per month *4. Prior to the launch of the service, pre-verification was conducted at stores operated by Chubu Yakuhin Co., Ltd., a group company of Valor Holdings Co., Ltd., which operates drugstores mainly in the Chubu region. %*5, demonstrating high efficacy. In the future, in addition to "predicting the number of customers coming to the store", we aim to develop a product demand prediction function, and plan to add functions such as inventory ordering and shift creation.

In Japan, 5.7 million tons*6 of food is wasted annually, and it has become a major social problem. In addition, especially in the retail and restaurant industries, securing human resources has become an issue due to the high turnover rate, changes in consumption behavior due to the impact of the new coronavirus infection, and the difficulty of foreign workers coming to Japan.

"Sakimir" contributes to reducing food loss and optimizing staff allocation by adjusting the number of product orders and work shifts according to the predicted number of customers visiting the store. It can be used when considering measures to improve sales, such as distribution. Through "Sakimir", SoftBank and the Japan Weather Association will support the retail and restaurant industries from various aspects such as operational efficiency and sales promotion by utilizing technologies such as data and AI to promote DX (digital transformation). . We also support the achievement of SDGs (Sustainable Development Goals) by contributing to the reduction of food loss and the improvement of productivity.

1. Highly accurate predictions using various weather data and unique AI algorithms Weather data such as temperature, solar radiation, wind speed, precipitation, snowfall, humidity, and weather held by the Japan Weather Association will be used. In addition, data scientists at SoftBank and weather analysis data scientists at the Japan Weather Association jointly developed AI algorithms that make use of their respective specialized skills, enabling highly accurate analysis and forecasting.

2. Predicting people's behavior by utilizing statistical data on the flow of people Based on the location information data of tens of millions of terminals obtained from SoftBank's mobile phone base stations, the statistical data on the movement of people expanded to estimate Japan's population of approximately 120 million. to grasp trends in the flow of people in the commercial area around the store. This makes it possible to make predictions that do not rely solely on past visit results, and even if people's behavior changes drastically in a short period of time due to the spread of new coronavirus infections, etc., predictions can be made based on the latest data on people flow. I can. Location information data will be anonymized so that individuals cannot be identified, and statistically processed data will be used.

3. Easy-to-introduce pricing system In general, services that use AI-based data analysis to predict the number of customers and product demand tend to be expensive to implement if they are customized in detail to meet the needs of the company in order to improve accuracy. I have. As a cloud-based, general-purpose model, “Sakimir” offers a “customer forecast” based on highly accurate data for a monthly fee of 5,390 yen (including tax) per store, making it easier to introduce. .

https://www.softbank.jp/biz/services/analytics/sakimiru/

*At the start of the service, it is a prerequisite for the introduction that the customer's system and "Sakimir" can be API-linked due to the specifications. In the future, we plan to relax the introduction conditions by continuing to develop the service so that a wide range of customers can introduce it. https://tm.softbank.jp/form/contact/alliance/index.php

*1 Forecast is made for each day for 14 days. At the start of the service, API linkage with the customer's system is required for introduction. *2 Population statistics data is anonymized and statistically processed so that individuals cannot be identified. *3 This service is provided using statistically processed data, and does not use data that can identify individuals. *4 A separate contract processing fee of 3,300 yen (including tax) per store is required. The minimum usage period is 12 months including the billing start month (the month following the usage start date). If you cancel during the minimum usage period, you will be charged a penalty (monthly usage fee x number of months remaining until the end of the minimum usage period). *5 Average forecast accuracy is calculated as "1-MAPE (average absolute error rate)". It was calculated from the results of preliminary verification conducted by Chubu Yakuhin Co., Ltd., and does not provide the same accuracy for all companies. * Verification period: March 2021, number of target stores: 10 * 6 Ministry of Agriculture, Forestry and Fisheries estimate for FY2019

•SoftBank and the SoftBank name and logo are registered trademarks or trademarks of SoftBank Group Corp. in Japan and other countries. •Other company names and product/service names mentioned in this press release are registered trademarks or trademarks of their respective companies.

PDF download: Started providing “Sakimir”, an AI demand forecasting service for the retail and restaurant industries that utilizes people flow and weather data