FUKUOKA -- An initiative in Japan to cut food waste using artificial intelligence-based correlation analysis between weather information and consumers' purchasing data to forecast demand for certain products has proven effective as some stores have seen the rates of discarded items drop and sales increase.
The trial was launched by the Japan Weather Association and the Fukuoka Municipal Government along with retailers in this southwestern Japan city. The weather body has been offering a demand forecast service, in which AI predicts sales of at least 660 items sold at stores based on a series of data including weather, what temperatures it "feels like," information circulating on social media and past sales data. The sales forecast is shown in seven stages, from "the very best time to sell" to "the worst time."
In the experiment run in the city of Fukuoka, meteorological factors in each of the city's wards as well as local specialties such as champon and udon noodles, among other variables, are added to the analysis. The program kicked off in May 2020 and will run through March 2021, with 10 stores from eight different businesses, including supermarkets and a bakery, participating in the trial.
The local business Tojin Bakery's Poem main store in the city's Chuo Ward, one of the participants in the program, saw about a 15% drop in the rates of products being discarded and a 12.3% increase in its sales proceeds in September and October compared to a two-month period before it started using the service.
The bakery's executive said, "When the rates of discarded products go up, our profit falls. It's nice to be able to see the numerical data for something that we've heretofore depended on our gut feelings (when it came to demand forecast) based on years of experience. Watching the forecast, we have been baking less in the morning and are now able to offer freshly baked goods around noon and in the early evening."
(Japanese original by Sayo Kato, Fukuoka News Department)