Analysis of the Effectiveness of Machine Learning Methods in Predicting Wheat Yields

Crop forecasting is one of the most important planning stages in the field of agricultural development and in economic planning in general. This article is devoted to forecasting wheat yields using machine learning methods. The feature of the article is forecasting using several approaches: machine learning methods and a combination of machine learning methods and AGROTOOL software. Four machine learning methods were used in the study: Random Forest, Ridge Regression, LASSO and XGBoost. The listed machine learning methods were chosen taking into account the characteristics of the data used in predicting wheat yield, namely, the multicollinearity of variables. In this regard, the random forest method was considered using the principal component method and without it. Based on the results of numerical experiments, it can be concluded that the Extreme Gradient Boosting method gave the best results, both as a single method and in combination with the AGROTOOL program.