Dual Stacking for Rapid Crop Yield Prediction Using LSTM-Based Analytical Deep Learning

Abstract

This paper presents a novel approach to rapid crop yield prediction through the integration of Dual Stacking and Long Short-Term Memory (LSTM)-based Analytical Deep Learning (ADL). The proposed DSA-LSTM framework leverages both short-term and long-term memory variables to efficiently forecast agricultural output. By utilizing Dual Stacking as a temporary repository during the Set Covering Approximation, the model enhances the accuracy and speed of yield prediction, addressing the complexities of agricultural data. The research explores various aspects of crop productivity, including environmental and soil factors, and applies advanced data mining techniques using WEKA to optimize attribute selection and clustering. Experimental results from the Hissar district in Haryana show the potential of this method for accurate yield forecasting. The study also emphasizes the importance of proper data mapping and fact-table management in enhancing prediction reliability. This methodology provides a scalable and adaptable solution for real-time agricultural decision-making, supporting stakeholders in improving crop management and productivity. The future scope of the research includes incorporating additional environmental, economic, and market factors for more precise predictions.