Machine learning-driven premium estimation in crop insurance: leveraging climate and commodity market big data

Koprivica, Marija and Bradić, Kristina and Gligorijević, Marija and Kočović De Santo, Milica (2026) Machine learning-driven premium estimation in crop insurance: leveraging climate and commodity market big data. Ekonomika poljoprivrede, 73 (3). pp. 871-882. ISSN 0352-3462

Text EP 3-2026lq (1) - converted.pdf - Published Version Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (461kB)

Abstract

This study proposes a machine learning-driven framework for premium estimation in crop insurance against drought risk, integrating climatic and commodity market data. The methodology combines a Logistic Regression model for predicting drought probability with an Extreme Gradient Boosting model for forecasting commodity prices and crop yields, using regional datasets from Vojvodina and Šumadija. The results show strong predictive performance across regions and metrics. Findings confirm that drought occurrence, yield and price volatility are driven by climatic and spatial heterogeneity, underscoring the need for region-specific premium setting. The study demonstrates the potential of machine learning techniques to support more risk-sensitive and actuarially consistent pricing in agricultural insurance.

Actions (login required)

View Item