Crop Yield Prediction Using Stacking Ensemble Model

This study focuses on enhancing crop production prediction under varying climate conditions using data from FAOSTAT and the World Data Bank spanning from 1990 to 2014. Our novel approach involves stacking two high-performing models to leverage their strengths and mitigate weaknesses. Extra Trees and Random Forest emerged as the top-performing methods, serving as base-learners, while an Elastic Net regularizer (combining L_1 and L_2) acts as the meta-learner. This ensemble effectively yields accurate predictions for the dataset. Our model surpasses standard regression metrics, including Coefficient of determination (r2), mean squared error (MSE), mean absolute error (MAE), and Adjusted r2, which accounts for less significant features. By integrating these techniques, we provide a robust solution for farmers and stakeholders, enabling them to make informed decisions in agronomy and crop selection, ultimately enhancing agricultural outcomes in the face of changing climatic conditions.