Published Online:March 2025
Product Name:The IUP Journal of Information Technology
Product Type:Article
Product Code:IJIT010325
DOI:10.71329/IUPJIT/2025.21.1.7-20
Author Name:Mythili R, Satya Naga Pavan Charith Potturi, Obili Sai Sudeep Reddy, Charishma Addagadda and Vadicherla Chris Marx Reddy
Availability:YES
Subject/Domain:Engineering
Download Format:PDF
Pages:7-20
Sustainable agriculture and food security rely heavily on precise crop selection based on environmental factors. Traditional crop recommendation systems, which primarily use static datasets and simple machine learning (ML) models, fail to dynamically adapt to varying soil, climate, and geographical conditions. This paper presents a Graph convolutional network (GCN)-based crop recommendation system, leveraging deep learning (DL) to model spatial and temporal relationships among agricultural variables such as soil quality, weather patterns, and historical crop performance. Unlike conventional methods, GCNs enable the incorporation of complex interdependencies between these factors by structuring them as graph-based data representations. The proposed system processes multisource agricultural data through nodebased learning, enhancing contextual understanding and adaptability to localized conditions. The performance evaluation against traditional ML models—random forest (RF) and support vector machine (SVM)—demonstrates that the GCN model achieves a superior accuracy of 92.8%, precision of 91.2%, and recall of 90.7%, significantly outperforming baseline methods. The results highlight the potential of GCNs in precision agriculture, offering dynamic and high-accuracy crop recommendations that improve agricultural efficiency and sustainability.
Agriculture is fundamental to food security and economic growth, but crop selection remains a critical challenge due to climate variability, soil degradation, and inefficient resource management (Doshi et al., 2018).