Efficient extended Kalman multi-head boosting forest for prediction of nitrogen demand from rice plant using SPAD meter data
Year: 2026
Type: article
Abstract: With rice providing millions of people with a livelihood and feeding over half of the world’s population, agriculture is essential to both global food security and economic stability, but several challenges limit its prediction accuracy. First, ineffective pre-processing of chlorophyll content variations across growth stages reduces the model’s predictive accuracy for nitrogen content. Multicollinearity in regression models complicates accurate nitrogen demand extraction from chlorophyll data. Moreover, existing models fail to account for epigenetic variability, limiting their accuracy in predicting nitrogen demand from Soil Plant Analysis Development (SPAD) chlorophyll data. Thus, novel Extended Kalman Sparse Ridge Stepwise Multi-Head Random Boosting Forest has been developed in this work, which uses SPAD meters to estimate nitrogen levels from chlorophyll content. Therefore, first, an Extended Kalman Sparse Principal Space Model is used to filter out the variations in chlorophyll and isolate nitrogen-induced changes from growth stage fluctuations. Then, Partial Least Ridge Stepwise Regression is employed to extract latent variables that capture relevant variance, stabilize coefficient estimates, thus ensuring reliable identification of variables that truly influence nitrogen demand. Finally, a Multi-head scaled dot Random Boosting Forest is used, which predicts nitrogen demands under environmental variability and epigenetic influences. The experimental results demonstrate that the proposed model achieves a higher accuracy of 0.99 and F1-score of 0.905 with a lower RMSE of 0.15%. Together, these models support optimized prediction of nitrogen demand value and allow accurate fertilization strategies, improving crop yield, and promoting sustainable agriculture.
Language: English
Topic: Smart Agriculture and AI
Subfield: Plant Science
Domain: Life Sciences
SDG: Zero hunger
Open Access status: gold
Estimated APC paid: $2,306
OpenAlex ID: W7140130490