Advances in AI for Smart Farming: A Review of Hybrid and Explainable Models for Crop Recommendation and Yield Prediction

Smart farming integrates Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL) to optimize crop productivity and resource use. By combining soil nutrient (NPK) and pH data with remote-sensing and IoT sensor inputs, modern AI systems generate location-specific recommendations and yield forecasts. This review systematically analyzes recent advances in ML and DL models including Random Forest, Support Vector Machine (SVM), Convolutional Neural Network (CNN), and Long Short-Term Memory (LSTM) architectures applied to crop recommendation and yield prediction. It highlights how hybrid and explainable models improve prediction accuracy and interpretability compared with conventional techniques. The study synthesizes findings from 2016–2025 literature to identify three persistent challenges: data heterogeneity, limited scalability, and lack of model transparency. Emerging approaches such as federated learning, edge computing, and explainable AI are discussed as practical pathways toward interpretable, privacy-preserving smart farming systems. The paper contributes a comparative synthesis of current research trends and outlines design priorities for developing AI frameworks that are not only accurate but also transparent and deployable in real-world agricultural environments.