Harnessing 3D-CNN, GRU, and Attention Mechanisms for Next-Generation Crop Yield Forecasting

The deep learning model is applied to a detailed study on crop yield prediction especially relating to Andhra Pradesh in this paper. Precise prediction of crop yield is important for maximizing agricultural planning and resource utilization. In this research, an innovative deep learning model for crop yield prediction in Andhra Pradesh is developed to tackle the problem of merging various spatiotemporal agricultural data, such as climatic factors, soil characteristics, and crop traits. Conventional models, including Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN), were compared with a new hybrid architecture that integrates 3D Convolutional Neural Networks (3D-CNN), Gated Recurrent Units (GRU), and an attention mechanism. The new hybrid model outperformed the others, with a validation loss of 0.0102, a mean absolute error (MAE) of 0.077, and an R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> value of 0.654. These findings suggest that deep learning methods, and especially the hybrid method, can offer data-driven insights for policymakers and farmers to inform decisions to optimize agricultural productivity and sustainability.