Crop Green Energy Score Prediction Using LightGBM Based on Climate and Geographic Variables

Abstract

Agricultural sustainability and renewable energy integration have become critical challenges due to climate change, increasing energy demand, and environmental degradation. Assessing agricultural systems based solely on crop yield is no longer sufficient, as energy efficiency and environmental impact must also be considered. One emerging indicator is the Crop Green Energy Score, which reflects the potential of crops to support sustainable and low-carbon agricultural practices. However, accurately predicting this score remains difficult due to complex interactions between climate variability and geographic conditions. This study proposes a machine learning–based prediction framework using the Light Gradient Boosting Machine (LightGBM) to estimate Crop Green Energy Score from climate and geographic variables. The model integrates meteorological features, including temperature, precipitation, solar radiation, humidity, and wind speed, with geographic attributes such as elevation and spatial location. Comprehensive data preprocessing, feature engineering, and hyperparameter optimization were applied to enhance prediction performance. The main contribution of this research lies in the development of a robust and efficient predictive model specifically designed for crop-level green energy assessment. Model evaluation was conducted using Root Mean Square Error, Mean Absolute Error, and coefficient of determination metrics. The experimental results demonstrate that the proposed LightGBM model achieves high prediction accuracy and outperforms baseline methods. Future work will focus on incorporating remote sensing data, expanding spatial coverage, and developing real-time decision support systems to further strengthen sustainable agriculture and green energy planning.