International Research Journal of Education and Technology

Crop prediction is crucial for optimizing agricultural practices, improving food security, and ensuring sustainable farming. This research focuses on evaluating various machine learning models for accurate crop prediction based on climatic, soil, and environmental data. Several state-of-the-art algorithms are tested, including Logistic Regression, Decision Trees, Random Forest, Gradient Boosting, and Support Vector Machines, among others. These models are assessed using K-fold cross-validation to determine their predictive performance, with a particular focus on accuracy and reliability.

After evaluating multiple models, the top three performing ones are identified and selected for further enhancement. To improve the accuracy of crop prediction, a RNE (Random Forest, Naive Bayes, and Extra Trees) Stack Ensemble algorithm is employed, which combines the top three models to form a new, integrated predictive model. This algorithm leverages the complementary strengths of each base model, with a meta-learner to make final predictions. The integration of the models results in improved prediction accuracy compared to the individual models.