Scientific News Report
A recent systematic review published in African Scientific Reports examined the growing use of machine learning and deep learning for crop recommendation in precision agriculture. From 183 identified studies, 129 publications from 2020 to 2026 were selected using a PRISMA-based screening process. Random Forest and XGBoost emerged as leading approaches, while Support Vector Machines, Decision Trees, k-Nearest Neighbours, CNNs, and LSTMs were also widely applied. Common inputs included nitrogen, phosphorus, potassium, soil pH, weather conditions, and vegetation indices such as NDVI and EVI. Despite high reported predictive accuracy, the review identified important limitations involving static datasets, limited real-time deployment, poor cross-regional validation, weak multi-source data integration, and model interpretability. The authors recommend more scalable, explainable, and real-world-ready crop recommendation systems.
Read report