Utilizing Machine Learning to Enhance Agronomic Practices and Crop Yield Prediction

Agricultural productivity plays a crucial role in sustaining global food security, and predictive models are essential for optimizing crop yields and agronomic practices. This research explores the effectiveness of three prominent machine learning algorithms—Sequential 196 Minimal Optimization (SMO), Logistic Model Tree, and Decision Table—to predict crop yields accurately. The dataset, comprising 2200 records, is sourced from the Kaggle dataset repository. By employing a voting ensemble approach, our study harnesses the collective predictive strength of these models. Notably, the Logistic Model Tree algorithm emerged as the most promising candidate for crop yield prediction, achieving an exceptional accuracy of 99.15%. This result highlights the robustness of the Logistic Model Tree in identifying intricate patterns within the dataset, making it a compelling choice for predictive modelling in agricultural practices. In comparison, the Sequential Minimal Optimization algorithm demonstrated an impressive accuracy of 97.83%, showcasing its capability to capture complex relationships within the data. The Decision Table algorithm achieved an accuracy of 93.77%, positioning it as a strong yet slightly less effective model compared to the Logistic Model Tree. Our findings strongly advocate for the adoption of the Logistic Model Tree as the preferred model for crop yield prediction due to its superior accuracy.