Enhanced stock price prediction with optimized ensemble modeling using multi-source heterogeneous data: Integrating LSTM attention mechanism and multidimensional gray model

The prediction of stock prices is a complex task due to the influence of various factors, high noise, and nonlinearity. This paper focuses on addressing the challenges of low prediction accuracy and poor stability, which have been a key area of interest in academic research. We proposed an optimized ensemble model that combines an LSTM-based attention mechanism and a cyclic multidimensional gray model, utilizing multi-source heterogeneous data. Our results demonstrate that the ensemble model achieves improved prediction accuracy, exhibits a good fitting effect, and outperforms individual models. The ensemble model yields smaller Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Root Mean Square Error (RMSE) values compared to the LSTM-based attention mechanism model and the multidimensional gray model. Furthermore, the ensemble model shows enhanced coefficient of determination (R²). Comparative analysis with alternative models such as ARIMA, GRU, CNN, and CNN-GRU reveals that the ensemble model achieves significant advancements in prediction accuracy.