Fuel-driven hybrid deep learning model for forecasting finger millet prices

In today’s data-driven society, forecasting future trends plays a critical role. Most decisions are made based on weather, finance, and Market trends. Motivated by the need for accurate agricultural market forecasting, this research work focuses on crop price prediction for Finger Millet (Ragi), a staple cereal crop. This work presents a hybrid predictive framework. It integrates deep learning models such as Long Short-Term Memory (LSTM), 1-Dimensional Convolutional Neural Network (1D-CNN), and Gated Recurrent Unit (GRU) with the statistical Vector Auto Regression (VAR) time series model. Among these, a hybrid VAR-LSTM model is proposed to enhance prediction accuracy by leveraging both temporal dependencies and multivariate input parameters. It includes crop arrival data and fuel prices. The framework’s performance is evaluated using standard error metrics: Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE). The proposed model achieves a MAPE as low as 0.01 (1%), with MAE and RMSE values of 33.8 and 36.7, respectively, outperforming existing benchmark models. Further, LIME (Local Interpretable Model-agnostic Explanations) technique is applied to understand the parameter contributions. Comprehensive result analysis highlights the significant influence of Fuel parameters on crop price trends. It emphasizes the model’s practical relevance for agricultural stakeholders and policymakers.