Crop yield forecasting is increasingly important for farmers facing uncertainties such as varying rainfall patterns and fluctuating soil nutrient levels. This study proposes a data-driven system that utilizes historical crop data, soil nutrient parameters, and weather conditions to predict agricultural yield with improved accuracy. Four machine learning models—Linear Regression, Decision Tree, Random Forest, and Long Short-Term Memory (LSTM)—were evaluated to capture both static and time-series characteristics of the data. Among these, the Random Forest model demonstrated superior performance due to its ability to model complex interactions between soil nutrients, rainfall, and temperature. Additionally, K-means clustering was applied to categorize soil types, and SHAP (SHapley Additive exPlanations) analysis was used to interpret the contribution of individual features in the prediction process. The proposed system provides practical insights that can assist farmers, particularly in regions like Podili, in making informed decisions regarding irrigation and fertilizer usage, ultimately enhancing agricultural productivity.
Nakkina et al. (Mon,) studied this question.