Vol. 7, Issue 9, Part A (2025)

Climatic fluctuations and plant and animal pests are still a serious threat to sustainable agriculture and food security. The study presents a Risk-Aware Agricultural Decision Support System where ensemble machine learning models are used to estimate a crop failure in changes of climatic and pest situations. The system enhances high-resolution views of risk patterns by incorporating multisource data, including weather measures, soil quality, the level of pest infestation, and satellite pictures. Prediction accuracy is increased by combining ensemble algorithms such as Gradient Boosting Machines, Random Forest, and XGBoost and stropping them with a weighted stacking approach to help improve accuracy and help resist over fitting. The system is dynamic and can prewarn farmers and policy-makers of local agro-climatic patterns. It promotes explain ability by offering SHAP-based feature attribution and it also allows real-time decision making via cloud-based dashboards. The model is measured in different crops and areas and it demonstrates greater accuracy and generalizability as compared to individual base learners. The created framework provides an effective scheme of risk management in precision agriculture using predictive intelligence, thus supporting the optimization of resources, resilience, sustainability of the food system. The proposed system performed better than other methods having highest accuracy of 92.7% and F1-score of 88.4%.