Title: Predicting crop yields in the Caribbean from pesticide use and annual rainfall using decision tree regressors

Authors: Temidara Daniel Oyedotun

Addresses: School of the Nations, 41–42 New Market Street, Georgetown, Guyana

Abstract: With the global population increasing at an unprecedented rate, food insecurity remains a growing concern. Food security entails ensuring that all individuals have consistent access to sufficient, safe, and nutritious food to maintain a healthy life. Achieving this goal is increasingly challenging amidst rapid demographic expansion. One promising approach to mitigating food insecurity involves predicting crop yields to enhance agricultural productivity and optimise resource use. This study explores the application of machine learning models - specifically, decision tree regressors - to predict crop yields based on annual pesticide usage (in tonnes) and rainfall levels (in millimetres) in the Caribbean nations. Focusing on two staple crops in the region - rice and cassava - the study utilises national-level data on pesticide application, precipitation, and crop yields (in tonnes per hectare) across multiple years. The predictive models demonstrated strong performance, with R² scores ranging from 0.75 to 0.99, indicating a high correlation between predicted and actual yields. These findings suggest that pesticide use and rainfall are critical determinants of crop productivity and, when effectively managed, can significantly contribute to improved national food security.

Keywords: agriculture; cassava; climate variables; crop yield prediction; decision tree regressor; food security; rice.

DOI: 10.1504/IER.2025.150063

Interdisciplinary Environmental Review, 2025 Vol.24 No.4, pp.291 - 308

Received: 16 Jun 2025
Accepted: 01 Aug 2025
Published online: 28 Nov 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article