LSTM-Based Deep Learning for Crop Production Prediction With Synthetic Data

Abstract

The Agri-industry forms the backbone of the economy and livelihood. Hence, efficient planning on resources and ensuring a steady food supply is vital. This model discusses the challenges of accurately predicting crop yields influenced by multiple dynamic factors. Traditional models suffer with the complexity, thus leading to inaccurate predictions. Also, the availability of reliable training data is scarce, which poses an additional problem in training. Existing solutions range from traditional statistical models based on historical data to modern AI techniques. While these approaches are better than conventional methods, they are still unable to address data scarcity, non-linear interactions and the dynamic complexities. This model aims to overcome the limitations using long short-term memory (LSTM) and integrating synthetic data. LSTM is able to decipher complex patterns and synthetic data provides additional training samples that can enhance accuracy. The overall potential of this proposed solution can help mitigate food scarcity and strengthen sustainability.