An Enhanced Long Short-Term Memory Recurrent Neural Network Deep Learning Model for Potato Price Prediction

Abstract

Regarding the potato market, pricing fluctuations are a significant factor, and unfortunately, they cause many issues for producers and consumers. It happens to result in food insecurity and economic instability. This study brings in an advanced LSTM-RNN model built to predict potato prices, which might alleviate the mentioned challenges. We gathered a historical potato price database and other economic variables, normalized by the Z-score normalization method to ensure all the data was consistent and credible. The model’s effectiveness was benchmarked against five traditional machine learning models: we used K-nearest neighbor, random forest, support vector regressor, linear regression, and gradient boosting regressor to classify isolated households and determine their socioeconomic status. The empirical data implied that our proposed LSTM-RNN model was more efficient than all comparison models, leading to an R² value of 0.98. The paper not only substantiates the plausibility of applying deep learning to address the agricultural market prediction issue but also serves as a guideline noting the capabilities of the LSTM-RNN routine in improving the decision-making processes for the farmers participating in the sector. This model supports a sustainable food system and a balanced economy by bringing price stability integral to designing and implementing strategies to address food security.