Approach and model for forecasting winter wheat yield using machine learning

An analysis of the relevant subject area and scientific literature on the use of intelligent approaches for forecasting and planning activities in agriculture has been conducted. This analysis highlights the feasibility of employing machine learning to predict processes in agriculture. The purpose of this article is to develop a model for predicting winter wheat yields using historical data and machine learning algorithms, while taking into account the specific characteristics of processes and resource use in agriculture. The proposed forecasting approach for winter wheat yields relies on historical data and machine learning algorithms that consider the unique aspects of agricultural processes and the resources involved. The selection of an effective model for predicting winter wheat yield is based on a developed algorithm, which involves a systematic implementation of seven stages. To prepare the data, the authors utilized intelligent analysis algorithms that assess the relationships between various factors affecting winter wheat yield. With qualitatively prepared data, the research substantiates the model for predicting winter wheat yield by evaluating its accuracy indicators. Three algorithms were chosen for the study: least squares (OLS), gradient boosting (XGBoost), and linear regression with polynomial features. Separate models were created for each algorithm and compared based on quality indicators. The findings indicate that the best model is the gradient boosting (XGBoost) model, which demonstrated the lowest values across all quality metrics - MSE, RMSE, MAE, and R-squared. Future research should focus on the development of an intelligent information system for planning agricultural processes, which includes a module for forecasting winter wheat yields based on the validated model proposed in this study.