Email Address

info@canadianpoultryresearchnetwork.ca

Our Location

Across Canada

Follow Us :

Development of a random forest model for classification of Canadian egg farms according to life cycle environmental impacts

2022 CPRF,Environment and Management
Over the past ten years, the Canadian egg industry has exhibited significant growth [1], accompanied by large reductions in environmental impacts compared to historical levels [2]. While this trend towards lower environmental impacts per unit of eggs produced is promising, in light of the large contributions food systems make to many environmental impacts [3,4] as well as the growth of the Canadian population and it’s demand for eggs [5] it is clear that continued improvements are necessary to ensuring the long term sustainability of the Canadian egg industry. Key to sustainability improvement efforts is an in-depth understanding of the differences in environmental impacts across different egg farms operating in Canada, as well the specific farm characteristics and management practices that drive those differences so as to support mainstreaming sustainability best practices. Previous investigations have resulted in little clarity regarding these drivers [6]. To continue this investigation, here we have developed a pair of machine learning models capable of accurately classifying farms as low-, average-, and high-environmental impact relative to their peers based on 159 individual farm-level life cycle assessments (LCAs), the gold standard for benchmarking of environmental impacts. These models, known as random forests, are able to classify farms according to their total life cycle environmental impacts with %100 and ~85%, depending on the set of predictor variables used. These models indicate that housing system and manure management system are the most important predictors of relative impacts; however, in the absence of this information, percentage of eggs discarded on the farm, mortality rate, and lay cycle length were found to be the most important variables for predicting relative impact levels. This work represents the first application of a random forest classification model for classifying farms based on life cycle environmental impacts. Both developed models performed well with regards to classification accuracy, and could provide farmers with an easy way to estimate their life cycle environmental impacts relative to their peers. In the future, the methodological developments presented here may provide substantial value to the Canadian egg industry, as well as many other industrial sectors.
Tags :
farm classification,life cycle assessment,sustainability
Share This :

Presenter

PhD Student

University of British Columbia

Follow on: