Home / Current Issue / Paper 1718040
A Multi-Objective Intelligent Decision-Support System for Sustainable Fertilizer Management, Nitrogen Leaching Prediction, and Carbon Footprint Monitoring in Precision Agriculture
Subject area: Agriculture and Veterinary Sciences · Area of research: Precision Agriculture, Machine Learning
DOI: https://doi.org/10.64388/IREV9I11-1718040
Abstract
In modern precision agriculture, fertilizer management systems primarily focus on maximizing crop yield and profitability, often overlooking long-term environmental consequences such as nitrogen leaching and carbon emissions. This paper proposes a multi-objective intelligent decision-support system that integrates machine learning models including Random Forest, CatBoost, and TabNet with Explainable AI (SHAP) and IoT-based real-time monitoring. Experiments were conducted on a real agronomic dataset of 23,682 samples sourced from the Kaggle Playground Series S5E6 competition. Results demonstrate that Random Forest, CatBoost, and TabNet jointly achieved the highest nitrogen leaching classification accuracy of 95.55% with an F1-Score of 0.9354 and AUC-ROC of 0.9453 and 0.9471 respectively, substantially outperforming Logistic Regression (81.42%). For carbon footprint estimation, Linear Regression achieved the best R2 of 0.7811 with an RMSE of 3.0022 kgCO2-eq/ha, followed closely by CatBoost (R2 = 0.7726). SHAP-based feature importance analysis identified Nitrogen content (63.45%) and soil Moisture (30.37%) as the dominant predictors of leaching risk. The proposed framework provides a scalable, interpretable solution for real-time sustainability tracking that harmonises crop yield, economic profit, and environmental stewardship.
Keywords
Sustainable Agriculture, Precision Farming, Machine Learning, Nitrogen Leaching, Carbon Footprint, IoT, Random Forest, CatBoost, TabNet, Explainable AI (SHAP), GIS Spatial Mapping, Fertilizer Recommendation
How to cite this paper
@article{1718040,
author = {Cyden Nilesh Costa, Dr. Haripriya V.},
title = {A Multi-Objective Intelligent Decision-Support System for Sustainable Fertilizer Management, Nitrogen Leaching Prediction, and Carbon Footprint Monitoring in Precision Agriculture},
journal = {Iconic Research And Engineering Journals},
year = {2026},
volume = {9},
number = {11},
pages = {2929-2940},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1718040.pdf},
abstract = {In modern precision agriculture, fertilizer management systems primarily focus on maximizing crop yield and profitability, often overlooking long-term environmental consequences such as nitrogen leaching and carbon emissions. This paper proposes a multi-objective intelligent decision-support system that integrates machine learning models including Random Forest, CatBoost, and TabNet with Explainable AI (SHAP) and IoT-based real-time monitoring. Experiments were conducted on a real agronomic dataset of 23,682 samples sourced from the Kaggle Playground Series S5E6 competition. Results demonstrate that Random Forest, CatBoost, and TabNet jointly achieved the highest nitrogen leaching classification accuracy of 95.55% with an F1-Score of 0.9354 and AUC-ROC of 0.9453 and 0.9471 respectively, substantially outperforming Logistic Regression (81.42%). For carbon footprint estimation, Linear Regression achieved the best R2 of 0.7811 with an RMSE of 3.0022 kgCO2-eq/ha, followed closely by CatBoost (R2 = 0.7726). SHAP-based feature importance analysis identified Nitrogen content (63.45%) and soil Moisture (30.37%) as the dominant predictors of leaching risk. The proposed framework provides a scalable, interpretable solution for real-time sustainability tracking that harmonises crop yield, economic profit, and environmental stewardship.},
keywords = {Sustainable Agriculture, Precision Farming, Machine Learning, Nitrogen Leaching, Carbon Footprint, IoT, Random Forest, CatBoost, TabNet, Explainable AI (SHAP), GIS Spatial Mapping, Fertilizer Recommendation},
month = {May},
doi = {https://doi.org/10.64388/IREV9I11-1718040}
}