Algorithmic diversity reduces risks of fertilizer recommendation errors
Algorithmic Monoculture in Machine Learning-based Site-Specific Fertilizer Recommendation
Machine LearningComputers and Society
Summary
Using machine learning models to recommend fertilizers for farms can lead to similar mistakes across many farms, creating bigger problems. The authors show that relying on just one popular model—an algorithmic monoculture—may increase overall errors and costs. They propose using a mix of different models to reduce these shared mistakes, which can lower the total social loss in fertilizer use. Their work uses data from Moroccan cereal systems to demonstrate how this approach can improve recommendations while balancing accuracy and risk.
What this means in practice
- •For agricultural data teams: Design fertilizer recommendation systems that blend multiple machine learning models to reduce correlated prediction errors and lower overall social losses in farm decision-making.
- •For environmental risk managers: Assess the systemic risks of relying on a single algorithmic model for fertilizer advice and improve ecosystem sustainability by encouraging diverse model adoption.
Tested on one dataset.
Authors
Abdelghani Belgaid
Abstract
Machine learning recommendations are often evaluated one model at a time, despite deployment across interconnected decision makers. For continuous NPK fertilizer prescriptions, correlated errors can aggregate across farms and generate systemic costs. A model of shared and farm-specific recommendation errors links expected social loss to model adoption shares; the Herfindahl-Hirschman Index characterizes concentration effects when common model errors are independent and predictive risks are equal. A covariance-generalized formulation captures the dependence between model families. A two-model threshold characterizes when privately optimal monoculture is socially dominated, and a convex portfolio rule limits average private predictive regret. Nine model families across five Moroccan cereal systems provide a benchmark-based calibration. Under illustrative parameters and a 10% regret cap, accuracy-constrained diversification lowers modeled social loss in all five systems by 10.89-41.53% (mean 19.41%). Uniform allocation reduces modeled loss in this setting, but breaches the regret cap in every crop. Correlation sensitivity shows that highly dependent model errors attenuate the benefits of diversification. These counterfactual outcomes are not estimates of agronomic or monetary welfare; the framework identifies adoption architecture and error dependence as ecosystem-level evaluation variables.