Abstract
<title>Abstract</title> <p> Wheat ( <italic>Triticum aestivum</italic> L.) is one of the world’s most important cereal crops, making research aimed at enhancing yield and improving agronomic and physiological traits a key priority. This study examined genotype × environment (G×E) interactions across five distinct regions. The results revealed significant G×E effects for most traits, except for variable fluorescence, chlorophyll index, and chlorophyll <italic>a</italic> , which exhibited minimal interaction. Mean comparisons of G×E interactions for grain yield identified G6 and G2 in the Damavand region and G9 in Qazvin as the highest-performing genotypes. Polygon plot analysis across all regions indicated that G21, G14, G10, and G5 were the most stable and desirable genotypes, while genotype ranking based on ideal performance favored G2, G12, and G10. Eigenvalue decomposition indicated that the first ten principal components accounted for more than 84% of the total variance, underscoring strong patterns in trait variation. Collectively, analyses across the studied regions identified G2, G10, and G12 as the most promising genotypes for future breeding programs. In predictive modeling, the extreme gradient boosting (XGBoost) algorithm demonstrated superior performance compared to random forest (RF) and support vector machine (SVM) models. For the test dataset, XGBoost achieved <italic>R</italic> ² values of 0.652 (Karaj), 0.628 (Qazvin), and 0.698 (Damavand), along with RMSE values of 0.084, 0.067, and 0.114, respectively. The mean absolute percentage error (MAPE) remained consistently low (3.014%, 2.503%, and 2.046%), confirming the robustness of the model. In conclusion, this study identified high-performing wheat genotypes (G21, G14, G10, and G5) with stable adaptability across diverse environments, making them strong candidates for breeding programs. Furthermore, the successful implementation of machine learning (RF, SVM, and XGBoost) highlights their potential to enhance yield prediction and optimize genotype selection strategies in wheat improvement programs. </p>