Wheat crop yield optimization using ML techniques in data analytics paradigm

Wheat production forecasting is crucial for sustainable agricultural development, resource allocation, and food security improvement. In this paper, we present HEO-Wheat (Hybrid Ensemble Optimization) for better wheat yield prediction by employing state-of-the-art machine learning in a data analytics approach. Random Forest, XGBoost, Support Vector Regression, and Artificial Neural Networks are incorporated in the framework through a weighted ensemble model to improve predictive power and generalization. In this study, we prepared agricultural datasets including soil nutrients, climate factors, irrigation patterns, and yield history using normalization, feature engineering, and correlation-based feature selection. The performance of the model was evaluated using RMSE, MAE, R², and MAPE with K-fold cross-validation. Experimental results show that the accuracy of HEO-Wheat is significantly higher, with lower error margins and better variance explanation than various individual models. The framework also includes a yield optimization module that provides data-driven recommendations for irrigation and fertilizer management, leading to significant yield improvements. The results of this paper demonstrate a scalable and accurate solution for precision agriculture in wheat production systems and support knowledge-driven decision-making.