Predictive Analytics for Food Loss Prevention, a DTU Skylab FoodLab x MIT Denmark project led by Andrea McClave, explores the possibility to predict on-farm food loss through predictive analytics and quantitative models. It aims to arm farmers with decision making tools, to mobilize resources for food rescue, and to mitigate cropland expansion by better utilizing what’s already being grown.
Today, reliable data on, on-farm food loss is scarce and numerous variables impact farmer decisions for when to end harvest (Baker et al. 2019; Johnson et al. 2019). The need for on the fly decision making coupled with a lack of field data (Johnson et al. 2019), leads to potentially much greater on-farm food loss of fruits and vegetables and roots and tubers than past estimates, which generally hover around 20% (Gustavsson, Cederberg, and Sonesson 2011).
Farmers currently face last minute decision making scenarios when deciding whether or not to harvest crops in their fields (Johnson et al 2019). Further, they often do not accurately know how much of their marketable yield is getting left behind in the field despite being of marketable and/or edible quality (Baker et. al 2019).
This project works to bring data on the major variables impacting farmer harvest decisions together - e.g. crop prices, estimated yields, labor availability and cost, available buyers - to predict when edible crops are at risk of being lost in the field. It aims to better enable planning and decision making capabilities along the food supply chain so that edible food, and the resources utilized to grow that food, do not go to waste. In doing so, the project will bring together multiple stakeholders across farmers, processors, retailers, data scientists, researchers, and policy makers and build a platform for future innovation in mitigating food loss.
To date, the project team at DTU Skylab FoodLab x MIT Denmark has determined key variables for building a mathematical model, established relationships with several leading researchers and organizations working in the space, and outlined data collection needs moving forwards. The project has great potential but requires funding for future research and data access, and the DTU Skylab FoodLab is interested in further establishing partnerships and collaboration to tackle this challenge.
Waste and loss in the Danish food supply chain
Today, estimates of country level food loss and waste are high with 716,000 tons of avoidable food waste occurring throughout the Danish Food Supply Chain (Tonini, Brogaard, and Astrup 2017). An estimated 14% of this avoidable food waste comes from primary production (Tonini, Brogaard, and Astrup 2017).
Waste and loss in the U.S. foodsupply chain
In larger countries such as the U.S. there is an estimated 63 million tons of food waste, 16% of which occurs in primary production causing farmers 15 billion USD in lost economic value (ReFED 2016).
Specifically looking at fruit & vegetable loss, global estimates are high with up to ~20% est. of loss at the agricultural level (FAO 2012); however, recent studies (Nordic & U.S.) highlight historically low estimates of on-farm food loss and the need for more accurate data (Baker et al 2019; Johnson et al 2019; Tonini, Brogaard, and Astrup 2017).
In particular, two recent studies by Baker et al. (2019) and Johnson et al. (2019) shed light on just how low past estimates might be for certain crop types and geographies. The two studies use direct, in-field measurement to document and audit post-harvest loss by crop type. Both studies capture how marketable, edible, inedible produce is left in the field after harvest (Baker et al 2019) and further the understanding of how much loss occurs in each category.
A recent report by the FAO, broadly defines food waste and loss as the “decrease in quantity or quality of food along the food supply chain” (FAO 2019). Food waste and loss across the supply chain create global environmental, social, and financial challenges. Food that never gets consumed utilizes resources such as water, fertilizer, and cropland (ReFED 2016), exacerbating agriculture’s environmental impact on the planet, and bringing into question moral and social obligations in combating global hunger and malnutrition.
Currently, the FAO and the UN are developing two frameworks to monitor and measure food loss and food waste as two separate entities, which also “aligns with the distinction implicit in SDG Target 12.3.”(FAO 2019). These reports set forth the following definitions:
Food losses: occur along the food supply chain from harvest/slaughter/catch up to, but not including, the retail level.
Food waste: occur at the retail and consumption level.
Definition parameters: “Food diverted to other economic uses, such as animal feed, is not considered as quantitative food loss or waste. Similarly, inedible parts are not considered as food loss or waste.” (FAO 2019, xii–xiii)
Building on work by Baker et al. and Johnson et. al., this project focuses on farm-level food loss, defined as crops that are left in the field and not harvested (Johnson 2018a). In-field loss of marketable and edible crops often occurs when produce does not meet market standards or when market prices are too low to justify harvesting (Baker et al).
Note, that the recent FAO 2019 calculations of food loss do not automatically include food loss at time of harvest. Their definition and calculations begin post-harvest (FAO 2019).
