Purpose. The purpose of the study is to model the yield of grain and legume crops in the Odessa region based on a complex of agro- and meteorological factors using regularized regression methods (Ridge, Lasso, ElasticNet). Methodology / approach. The article applies economic and mathematical analysis using official statistics for 1995–2024. 11 independent variables were selected for modeling, reflecting agro-technological and climatic factors. The assessment was carried out using classical linear regression and its regularized modifications. Multicollinearity diagnostics, cross-validation, and analysis of statistical significance of coefficients were performed. Results. It is shown that models with regularization provide higher accuracy of yield forecasting compared to classical regression. The best results were demonstrated by Ridge regression (R² = 0.606; RMSE = 4.46), which revealed key positive factors: crop area, rainfall, application of mineral fertilizers and pesticides, hydrothermal coefficient. The deficit of air saturation with moisture and wind speed had a negative impact on yield. Originality / scientific novelty. The scientific novelty lies in the combination of agronomic and meteorological indicators within the framework of regularized models, which allows taking into account multicollinearity and at the same time identifying the most significant yield factors. The results obtained deepen knowledge about the relationships between agrotechnological and climatic factors in the conditions of the Southern Steppe of Ukraine. Practical value / implications. The results of the study can be used by agricultural enterprises and management bodies to optimize fertilizer systems, plan crop areas, select adaptive varieties, and increase the resilience of agricultural production to climate risks. The proposed models provide tools for predicting yields and making management decisions in the field of agro-economics.
Kulyk et al. (Wed,) studied this question.