Japan Weather Association and NEC Collaborate to Reduce Food Waste Using Big Data

Image photo.

A new initiative in Japan is tackling the global issue of food loss and waste by attempting to eliminate mismatches between food supply and demand. This "Supply and Demand Optimization Platform" can share and utilize crucial information in the entire value chain.

The Japan Weather Association (JWA) and NEC Corp., a leading Japanese electronics company, announced on February 28, 2018, that they would collaborate to optimize supply and demand in entire value chains, including manufacturing, distribution, logistics and sales in various industries and sectors.

To begin with, they applied their platform to the food value chain. By optimizing the supply and demand of not only individual industries but the entire value chain by providing user companies with demand forecasting data, inventory information and sales results, they aim to help address the social issue of food loss and waste.

The platform utilizes a technology called "Heterogeneous Mixture Learning Technology" developed by NEC. Besides automatically discovering specific patterns based on correlations in big data, it can switch reference patterns depending on the analyzed data. With traditional machine learning it is generally difficult to discover more than a single pattern, but this new platform has high-precision prediction and anomaly detection even in data with varying patterns.

One key feature of this particular technology is that it clearly shows the underlying explanations for the forecasts it makes, and this facilitates smoother decision-making and implementation. With its solid track record in consulting on commodity demand forecasting, JWA will provide meteorological data for the platform's data analysis, and also offer commodity demand forecasting services using the data.

To boost the platform's accuracy, the partners have been conducting a demonstration trial for beverages since January 2018. Besides demand forecasting for the retail industry, they have already begun to see a significant improvement in the accuracy of product demand forecasting for manufacturers by utilizing information directly associated with consumption, including retail sales data, and weather and event information.

The platform is expected to help reduce food loss and waste, through production planning and order planning based on highly accurate demand forecasts, and adjusting the supply-demand balance based on fluctuations in demand forecasts. The partners are also considering future development of the platform beyond demand forecasting, by linking with operational systems such as supply-and-demand planning, production planning, order planning, and inventory management.