Answers are grounded in this paper and cite their sources.
Rice yield optimization under intensified nitrogen (N) management requires understanding of complex interactions among nutrient inputs, crop physiological traits and environmental factors, which are often inadequately captured by conventional statistical approaches. Our field study evaluates five N and four zinc (Zn) fertilization strategies, including nano-urea sprays to optimize yields while reducing N inputs. Although growth attributes, yielding traits, and nutrient uptake are critical for yield prediction, yet their integration into machine learning (ML) models remains limited. To address this, ML algorithm models like linear regression, support vector regression (SVR), k-nearest neighbors (kNN), gradient boosting ( gbm ), and random forest ( rf ) were evaluated using statistical performance indicators with R 2 , root mean square error (RMSE), adjusted RMSE, and mean absolute error (MAE). Among the tested models, kNN showed the lowest predictive performance, whereas the RF model performed best. The RF model substantially outperformed all competing algorithms, achieving an MAE of 0.06 and explaining 97% and 96% of yield variability across the two seasons, while demonstrating strong generalization and resistance to over fitting following systematic hyper parameter tuning. GHG emissions were estimated using the CCAFS-MOT tool, with RN 97.5 kg ha −1 + NU having 6.55% lower CO 2 -equivalent emissions per hectare and 3.22–4.62% lower emission intensity than the recommended RN 130 kg ha −1 , without compromising grain yield. These findings establish RF modelling as a reliable framework for site-specific rice yield prediction, demonstrating that nano-urea-based nitrogen reduction combined with ML offers a scalable, evidence-based pathway for simultaneously improving productivity and reducing environmental impact in irrigated rice systems.
No takes yet. Share an insight, caveat, or question.
Reddy et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: