Data Mining and Machine Learning Algorithms for Optimizing Maize Yield Forecasting in Central Europe

Artificial intelligence, specifically machine learning (ML), serves as a valuable tool for decision support in crop management under ongoing climate change. However, ML implementation to predict maize yield is still limited Central Europe, especially Hungary. In this context, we assessed the performance of four algorithms (Bagging (BG), Decision Table (DT), Random Forest (RF) and Neural Network-Multi Layer Perceptron (ANN-MLP)) predicting based on different input scenarios. The collected data included both agricultural (production (PROD) (ton) cropped area (AREA) (ha)) (annual mean temperature °C (Tmean), precipitation (PRCP) (mm), rainy days (RD), frosty (FD) hot (HD)). This research adopted scenarios, follows: SC1: AREA+ PROD+ Tmean+ PRCP+ RD+ FD+ HD; SC2: PROD; SC3: SC4: PRCP. training stage, ANN-MLP-SC1 ANN-MLP-SC4 outperformed other algorithms; correlation coefficient (r) was 0.99 both, while root squared errors (RMSEs) were 107.9 (ANN-MLP-SC1) 110.7 (ANN-MLP-SC4). testing phase, had highest r value (0.96), followed by (0.94) RF-SC2 (0.94). 10-fold cross validation also revealed that have performance. We further evaluated regional scale (Budapest). succeeded reaching high-performance standard (r = 0.98, relative absolute error 21.87%, 20.4399% RMSE 423.23). promotes use ANN an efficient yield, which could be highly beneficial planners makers developing sustainable plans management.