AEAPIM-CC: A Cloud-Enabled Integrated Model for Agricultural Economic Forecasting via Feature-Matrix Analysis and ARIMAX

Abstract

Considering the sharp growth in agricultural economic data and the shortcomings of current analytical methods, this article proposes an Agricultural Economic Analysis and Prediction Integrated Model based on Cloud Computing (AEAPIM-CC). The model employs an enhanced mutual information (IMI) method with conditional input filtering to facilitate feature selection and eliminate data redundancy. For measuring internal relationships within the data, an association analysis algorithm utilizing matrix decomposition is employed. For time series forecasting, an augmented autoregressive integrated moving average with exogenous inputs (ARIMAX) model is applied, which effectively captures both autoregressive patterns and the effect of external influences. AEAPIM-CC is tested with the Global Agricultural Economic Database (GAED) and compared against some linear regression (MLR), support vector machine (SVM), grey prediction GM (1, 1), and autoregressive (AR) models. Compared to the best-performing baseline (AR), AEAPIM-CC achieves an RMSE reduction of 0.99, MAE reduction of 3.70, MAPE reduction of 3.32%, and R² improvement of 0.15—demonstrating substantial gains across all performance metrics. These results demonstrate significant improvements compared to classical models in all indicators. This research not only promotes cloud computing applications in agricultural economic prediction but also provides strong support for decision-making in agricultural enterprises and government departments, thereby promoting the more scientific and sustainable development of the farm economy.