Artificial Intelligence-based Tools for Fresh Produce Procurement Price Decisions as Applied to Canadian Distribution Centers

点击翻译按钮获取中文摘要

Loblaw Companies Limited (LCL), employing about 200,000 Canadians [$45B annual total revenue], supplies from Waterloo Distribution Center (DC) all fresh produce (FP) to their stores in South Western Ontario that extend from west of highway 427 to the City of Windsor. LCL's food DCs are very large warehouses with many temperature zones for storing FP and are responsible for ordering and distributing to various store locations. Ordering depends mostly on the demand for FP and is expected to meet demand but procurement prices are also affected by expected crop yields. Adequately timed and priced orders bring financial benefits to LCL and at the same time minimizes waste. For example, U.S. estimates an annual loss of 5.9 and 6.1 billion pounds for fresh fruit and fresh vegetables, respectively [1]. Prices, however, depend on many factors that are affected by the diverse regions from which the FP is procured and affected by high uncertainty due to environmental and socio-economic effects such as income, labor and other trade issues. These factors are becoming even more uncertain from globalization and climate change, and hard to predict, making decisions on FP procurement prices and quantities an extremely challenging task. This important task is currently done based on immediate past prices of the same produce, which is too simplistic. Because of the great volume of transactions and monetary value (over $25B just for Loblaws FP), an improvement of even a micro cent in each FP transaction transfers to benefits of hundreds of millions of dollars each year for Canada. Additionally, fair prices bring prosperity all around from growers to consumers and to the general welfare of the involving regions. The urgent need and support of the LCL and the well-established research record of investigators at the University of Waterloo (UW) have provided an opportunity to develop, test, and employ advanced artificial intelligence (AI) and machine learning (ML) tools. This will result in improved FP "fair" procurement price offers using models that combine ML methods predicting crop yields with economic models. This novel approach will provide increased benefits to both FP industry and Canadian society.