<![CDATA[Accurate varietal identification of rice grains is crucial for quality assessment and data-driven decision-making in agricultural informatics. This study aims to comparatively eval-uate seven machine learning algorithms for morphology-based classification of Cammeo and Osmancik rice varieties and to identify the most suitable model for structured numerical grain-feature data. Using a dataset of 3,810 instances with seven image-derived morpho-logical features, a systematic comparison was conducted across Logistic Regression, Multi-Layer Perceptron (MLP), Support Vector Machine (SVM), Decision Tree, Random Forest, Naive Bayes, and k-Nearest Neighbors. The models were evaluated based on classification quality and computational efficiency. Results show that MLP achieved the highest overall predictive performance with an accuracy of 93.03% and an F1-score of 94.17%. However, when balancing accuracy against computational overhead, SVM emerged as the optimal” sweet spot” for industrial implementation, offering a competitive 92.50% accuracy with a 93-fold reduction in execution time compared to MLP. Naive Bayes demonstrated the fastest computational runtime (0.0022 seconds total). The study identifies a distinct trade-off between predictive quality and runtime efficiency, recommending MLP for high-fidelity research and SVM for real-time agricultural informatics applications.]]>
Copyrights © 2026
