Next-gen agriculture: integrating AI and XAI for precision crop yield predictions

Area of Science:

• Agricultural Science
• Environmental Science
• Computer Science

Background:

• Climate change impacts global food security through altered precipitation and extreme weather events, affecting crop yields and farmer economics.
• Understanding the complex interplay between climate and agronomic factors is crucial for developing effective adaptation strategies.

Purpose of the Study:

• To predict crop yields and assess climate change impacts on agriculture using AI and XAI.
• To provide a transparent framework for agricultural decision-making and policy development.

Main Methods:

• Exploratory Data Analysis (EDA) to identify key climatic and agronomic factors influencing crop yields.
• Application of advanced regression models (Decision Tree, Random Forest, LightGBM) for predictive performance evaluation.
• Utilization of Explainable AI (XAI) techniques, including SHAP and LIME, for model interpretability.

Main Results:

• Temperature identified as the most critical factor affecting crop yields, with significant rainfall and macronutrient interactions.
• High predictive accuracy achieved by regression models (R²=0.92, MSE=0.02, MAE=0.015).
• XAI techniques provided actionable insights into feature importance, enhancing understanding of model predictions.

Conclusions:

• AI and XAI offer a robust and transparent framework for mitigating climate change effects on agriculture.
• The study's findings support scalable applications in precision farming and climate policy development.
• Enhanced interpretability of AI models facilitates informed agricultural decision-making and adaptation planning.