• Jiji


The city of Fukuoka, jointly with the Japan Weather Association, is conducting an experiment to reduce food waste using artificial intelligence.

In the experiment, AI is used to predict sales of products in line with weather conditions, allowing stores to adjust their order and production volumes.

Participating stores were able to reduce waste and boost sales in the fiscal year that ended in March.

The experiment uses the JWA’s weather-based demand prediction service, which analyzes mainly weather conditions, temperatures, social media posts and past retail sales data to predict demand for more than 660 products, including fresh food and prepared food, in seven stages.

In the experiment last fiscal year, six of the eight participating companies in the city saw their food waste decline, while seven logged increased sales.

Tojin Bakery logged a drop of some 15% in food waste in September to October 2020, compared with July to August the same year, before the experiment began, while sales increased about 12%.

Using the prediction service, the bakery reduced production on days in which rain was expected and prepared freshly baked bread when customer traffic was expected to rise.

“It was helpful to have concrete numbers to show what we previously had relied on intuition for,” Kei Kuwano, managing director at the company, said. “We were also able to change staff members’ mindsets in an effort to reduce waste.”

The prediction service is expected to undergo upgrades, such as making it easier for stores specializing in a small number of products to use and adding a notification feature that alerts users to sudden changes in weather. The city is continuing the experiment in the current fiscal year.

“Reducing product waste is rewarding for the store and can help cut costs and lost sales opportunities,” an official of the city government section in charge of promoting the reduction of industrial waste said.

According to Japan’s agriculture ministry and other sources, a total of 6 million tons of food waste that could have been eaten was estimated to have been produced in fiscal 2018, with 3.24 million tons being produced by food-related businesses and 2.76 million tons coming from households.

In a time of both misinformation and too much information, quality journalism is more crucial than ever.
By subscribing, you can help us get the story right.