Japan Weather Association
Contributing to the Resolution of Social Challenges such as Food Waste Losses with Highly Accurate Commodity Demand Forecasting
Forecasting Commodity Demand Based on Big Data Including Weather Information
The Japan Weather Association (JWA) has been involved in various businesses related to disaster prevention, environment and energy, and information services with a focus on the provision of meteorological information. For about 20 years, JWA has been working on "weather merchandising" that utilizes weather and meteorological information for merchandising. Currently, JWA provides clients in the retail and manufacturing industries with a wide array of support including demand forecasting and production planning.
JWA wanted to realize more detailed commodity demand forecasting based on big data with diverse information such as commodities, locations, calendar, and POS data by utilizing weather forecasting data that has become 30% more accurate in the past 15 years. Demand forecasting covers a wide variety of commodities including apparel, cosmetics, convenience goods in addition to highly seasonal commodities such as ice cream and beverages. Utilization of AI is essential for analysis of big data and demand forecasting for a wide variety of commodities.
Realizing Highly Accurate Demand Forecasting by Utilizing NEC's AI
In order to utilize AI to realize a higher level of accuracy in demand forecasting, we looked into the Heterogeneous Mixture Learning Technology, one of NEC's cutting-edge AI technologies from their NEC the WISE lineup. The Heterogeneous Mixture Learning Technology is a type of machine learning that discovers specific patterns based on association and automatically creates prediction formulas. It can provide rapid, high-precision predictions based on mixed types of data. Around the summer of 2017, we started to consider a collaboration that involved combining JWA’s wealth of weather data and consultation expertise in demand forecasting with NEC's AI. Currently, we are working on demonstration experiments to develop an analysis system for optimizing supply and demand, number of deliveries, and delivery routes for beverage vending machines.
“In NEC's Heterogeneous Mixture Learning Technology, forecasts are based on a white box approach. This is one of the major differences from other types of machine learning. It clearly shows you the reasons and causes for forecasts, which you can explain to your customer in a comprehensible way.”
Toward Optimization of Supply and Demand by Connecting Manufacturing, Logistics, and Distribution
It has been reported that one third of all industries in the world are exposed to some form of weather-related risk. The Japan Weather Association aims to contribute to the resolution of a wide variety of challenges faced by our clients by providing highly accurate demand forecasting data and shrewd consultation. With highly accurate demand forecasting, we support the manufacturing, logistics, and retail industries with serious staffing shortage to help them reduce stockouts and food waste as well as increase sales and profits through appropriate promotional efforts.
"Our intention to optimize the entire supply chain by reducing various losses resonates with the Supply and Demand Optimization Platform that NEC is working on. We have a future collaboration in mind toward the creation of an optimization system for supply and demand planning based on weather data of the Japan Weather Association and demand forecasting with NEC's AI technologies." (Honma)
NEC will continue to contribute to the resolution of social challenges such as food disposal losses and manpower shortage and the creation of a sustainable society through effective utilization of resources by realizing a common platform system for total optimization of supply and demand across manufacturing, logistics, and distribution.
Insight from NEC Frontline Staff
Hirokazu Hira, Manager
Value Integration Department
Process Industries Solutions Division, NEC
In the future, it is important to establish a common foundation for optimization of the entire value chain across manufacturing, logistics, and distribution rather than dealing with each of them separately. Currently, NEC is working on the creation of the Supply and Demand Optimization Platform that combines accurate demand forecasting and a wide variety of causal data. We aim to solve social challenges by reducing losses in the entire process including production planning, delivery routes and number of deliveries, sales opportunities, food waste, etc.
Dan Fujiwara, Chief
Food and Consumer Goods Industries Integration Department
Process Industries Solutions Division, NEC
NEC's Heterogeneous Mixture Learning Technology used in this collaboration can visualize the reasons and grounds for forecasts as well as data on which forecasts are made, allowing you to make strategic decisions and carry out tasks smoothly. NEC's strength in the creation of the Supply and Demand Optimization Platform includes extensive expertise and knowledge in various industries we have been accumulating while providing solutions. We will continue to leverage our collective capabilities to develop solutions for challenges facing society and work toward achieving the UN Sustainable Development Goals (SDGs).