FEASIBILITY OF USING MACHINE LEARNING ALGORITHMS FOR YIELD PREDICTION OF CORN AND SUNFLOWER CROPS BASED ON SEEDING DATE

Abstract

In this research, our objective is to identify the relationship between the date of seeding and the production of corn and sunflower crops. We evaluated the feasibility of using prediction models on a dataset of annual average crop yields and information on plant phenology, from several states of the US. After performing data analysis and preprocessing, we trained a selection of regression models. The best results were obtained for corn using HistGradientRegressor and XGBRegressor with R2 = 0.969 for both algorithms and MAE% = 8.945%, respectively MAE% = 9.423%. These results demonstrate a good potential for the problem of yield prediction based on year, state, average plating day, and crop type. This model will be further used, combined with meteorological data, to build an agricultural crop prediction model. Received by the editors: 8 December 2022. 2010 Mathematics Subject Classification. 94A15, 94A99. 1998 CR Categories and Descriptors. H.1.1 [Information Systems]: Models and Principles – Systems and Information Theory; H.4.2 [Information Systems Applications ]: Types of Systems – Decision support I.2.1 [Artificial Intelligence]: Applications and Expert Systems – Medicine and science.

Keywords

• regression, yield prediction, seeding date, agriculture, XGBoostRegressor