CropCast: Harvesting the future with interfused machine learning and advanced stacking ensemble for precise crop prediction

Abstract

The Agro-Ecological (AE) zone plays a vital role in determining which crops are suitable for cultivation in specific areas of land. However, changing environmental and climate conditions have made it increasingly difficult for farmers to choose the right crops, resulting in both time and financial losses. Traditional research in this field has struggled with issues like inaccurate predictions, crop variability, diversification, and a high error rate. This research introduces the Inter-fused Machine Learning with Advanced Stacking Ensemble model (IML-ASE) as a solution to improve crop prediction accuracy using AE zone data. The primary goal is to develop the IML-ASE approach, leveraging information about agricultural, environmental, and soil conditions within AE zones to help farmers make knowledgeable decisions about crop selection. The process involves collecting data from crop recommendation datasets, preprocessing it, and feeding it into the proposed model. The advanced stacking model consists of multiple layers, with the first layer using various ensemble techniques as base learners, the second layer acting as a meta-learner, and the third layer serving as the fine learner. This research addresses the challenges faced in modern agriculture by providing farmers with more accurate crop predictions based on AE zone characteristics. Prediction performance is assessed by metrics like mean absolute error, mean square error, root mean square error, accuracy (97.1%), F1-score (97.09%), precision (97.03%), recall (97.12%), and specificity (100%), allowing for comparisons across predictions. This advancement aims to empower farmers to make informed decisions and effectively adapt to the everchanging agricultural landscape.