Forecasting the mechanical compaction influence on soybean yield using XGBoost-ANNOA

Soil compaction in agricultural fields caused by machinery operations is gradually becoming an important constraint to sustainable agricultural development.Predicting changes in crop yields under compacted environments and warning can help improve crop yields.However,relevant studies are lacking.The objective of this paper is to establish a prediction model for soybean yield changes in the mechanical compaction environment and to explore the predictive ability of the MLR,XGBoost and ANN.We proposed a two-step model to predict the crop yield changes based on the relationship among agricultural machinery operations,soil properties,and crop yield.For acquiring experimental data,we used three types of tractors(large,medium,and small)to complete the field compaction tests.The soybean yield changes model based on XGBoost-ANN hybrid approach has higher precision with R2 of 0.889,MAE of 1.47,and RMSE of 1.964.We also verify the effectiveness and robustness of the XGBoost-ANN model using the compaction data from the second year.Moreover,according to the results of the feature importance analysis,we give some suggestions for mitigate the effects of mechanical compaction.We demonstrate the feasibility of predicting changes in crop yields in compaction environments with good results and is important for preserving soil resources and enhancing crop productivity.