Enhancing Market Efficiency in ODOP's Jaggery Industry through Machine Learning Algorithms: Evidence from LSTM and Random Forest-Based Price Prediction

Strong forecasting models were necessary for market stability because of the high price volatility of the jaggery market in Muzaffarnagar, a major agricultural hub. In order to improve the accuracy of price forecasting, this study proposes a hybrid predictive framework that combines Random Forest (RF) regression with Long Short-Term Memory (LSTM) networks. Weekly jaggery prices and market arrivals were incorporated into time-series data from 2015 to 2023. Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R2 Score were used to build and compare three models: LSTM, RF, and a Hybrid LSTM-RF model. With a MAE of 30.83, RMSE of 27.45, and R2 of 0.9845, the Hybrid model demonstrated improved predictive ability over both LSTM and RF. Five-fold cross-validation was used to verify the model's robustness, and the Diebold-Mariano test verified that it was statistically superior to standalone models (p < 0.05). Jaggery prices for 2024 were then predicted using the improved hybrid model, which yielded important market patterns. The results demonstrate how effective AI-driven predictive analytics was in agricultural markets, especially when it comes to lowering pricing uncertainty and assisting stakeholders in making well-informed decisions. This work adds to the larger conversation on AI-driven market forecasting under ODOP programs and highlights the potential of hybrid machine learning approaches to increase market efficiency.