Mohan R N V Jagan, Rayanoothala Pravallika Sree, Sree R Praneetha
Department of Computer Science and Engineering, Sagi Rama Krishnam Raju Engineering College, Bhimavaram, India.
Department of Plant Pathology, MS Swaminathan School of Agriculture, Centurion University of Technology and Management, Odisha, India.
Front Plant Sci. 2025 Jan 8;15:1451607. doi: 10.3389/fpls.2024.1451607. eCollection 2024.
Climate change poses significant challenges to global food security by altering precipitation patterns and increasing the frequency of extreme weather events such as droughts, heatwaves, and floods. These phenomena directly affect agricultural productivity, leading to lower crop yields and economic losses for farmers. This study leverages Artificial Intelligence (AI) and Explainable Artificial Intelligence (XAI) techniques to predict crop yields and assess the impacts of climate change on agriculture, providing a novel approach to understanding complex interactions between climatic and agronomic factors. Using Exploratory Data Analysis (EDA), the study identifies temperature as the most critical factor influencing crop yields, with notable interactions observed between rainfall patterns and macronutrient levels. Advanced regression models, including Decision Tree Regressor, Random Forest Regressor, and LightGBM Regressor, achieved exceptional predictive performance, with R² scores reaching 0.92, mean squared errors as low as 0.02, and mean absolute errors of 0.015. Additionally, XAI techniques such as SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) enhanced the interpretability of the predictions, offering actionable insights into the relative importance of key features. These insights inform strategies for agricultural decision-making and climate adaptation. By integrating AI-driven predictions with XAI-based interpretability, this research presents a robust and transparent framework for mitigating the adverse effects of climate change on agriculture, emphasizing its potential for scalable application in precision farming and policy development.
气候变化通过改变降水模式以及增加干旱、热浪和洪水等极端天气事件的发生频率，对全球粮食安全构成重大挑战。这些现象直接影响农业生产力，导致作物产量降低，给农民造成经济损失。本研究利用人工智能（AI）和可解释人工智能（XAI）技术来预测作物产量，并评估气候变化对农业的影响，为理解气候和农艺因素之间的复杂相互作用提供了一种新方法。通过探索性数据分析（EDA），该研究确定温度是影响作物产量的最关键因素，同时观察到降雨模式和大量营养素水平之间存在显著相互作用。包括决策树回归器、随机森林回归器和LightGBM回归器在内的先进回归模型取得了出色的预测性能，R²分数达到0.92，均方误差低至0.02，平均绝对误差为0.015。此外，诸如SHAP（SHapley加法解释）和LIME（局部可解释模型无关解释）等XAI技术提高了预测的可解释性，提供了关于关键特征相对重要性的可操作见解。这些见解为农业决策和气候适应策略提供了依据。通过将人工智能驱动的预测与基于XAI的可解释性相结合，本研究提出了一个强大且透明的框架，以减轻气候变化对农业的不利影响，强调了其在精准农业和政策制定中可扩展应用的潜力。
