Machine Learning Approaches for Forecasting Raspberry Prices and Farm Profitability in Serbia

Abstract Price volatility and uncertain profitability remain critical challenges for Serbia’s raspberry sector, which is among the largest in the world and highly dependent on export markets. This study develops and evaluates machine learning (ML) approaches for forecasting raspberry prices and profitability, using farm-level production and economic data collected between 2020 and 2024. Several models, including multiple linear regression, random forest, gradient boosting, and long short-term memory (LSTM) neural networks, were trained and validated on historical data incorporating yields, input costs, prices, and cultivar characteristics. Forecast accuracy was assessed using mean absolute error (MAE) and root mean square error (RMSE). The results indicate that ensemble methods (random forest and gradient boosting) provided the most robust short-term price forecasts, whereas the LSTM achieved superior performance in capturing non-linear dynamics and seasonal fluctuations. The profitability predictions revealed that family labor costs and price variability were the strongest explanatory factors associated with the gross margin risk. Overall, the findings demonstrate that machine learning offers a valuable tool for anticipating market outcomes and managing economic risk in export-oriented horticulture. By integrating predictive analytics into farm-level and policy decision-making, Serbia’s raspberry producers can improve planning, stabilize income, and strengthen competitiveness in volatile international markets.