Sustainability assessment and modeling based on supervised machine learning techniques: The case for food consumption

Abstract

Sustainability of food consumption requires the understanding of multi-dimensional environmental, economic and social impacts using a holistic and integrated sustainability assessment and modeling framework. This article presents a novel method on the assessment and modeling of sustainability impacts of food consumption. First, sustainability impacts of food consumption categories are quantified using high sector resolution input-output tables of U.S. economy. Later, an integrated sustainability modeling framework based on two supervised machine-learning techniques such as k-means clustering and logistics regression is presented. The proposed framework involves five steps: (1) economic input-output life cycle sustainability assessment, (2) non-dimensional normalization, (3) sustainability performance evaluation, (4) centroid-based clustering analysis, and (5) sustainability impact modeling. The findings show that the supply chains of food production sectors are accounted for major environmental impacts with higher than 80% of portions for total carbon footprints. Animal slaughtering, rendering, and processing is found as the most dominant sector in most of the environmental impact categories. The logistic model results revealed an overall model accuracy equal to 91.67%. Furthermore, among all the environmental sustainability indicators, it has found that CO and SO2 are the most significant contributors. The results also show that 13.7% of the food and beverage sectors are clustered as high, in which the bread and bakery product manufacturing is the central sector. The large value of the variance (5.24) is attributed to the large total weighted impact value of the animal (except poultry) slaughtering, rendering, and processing cluster.