ARIMA-BIGRU Stock Forecast Model Based on Bayesian Optimization

The paper proposes a BAG(BO-ARIMA-BIGRU) model. Time series forecasting, as a key branch in the fields of machine learning and artificial intelligence, plays a very important role in areas such as finance, logistics, and climate prediction. The paper first elaborates on the related theories and methods of time series forecasting, including ARIMA model, BIGRU model, etc. Subsequently, the BAG model is constructed, which integrates the advantage of the ARIMA model in handling global information and the advantage of the BIGRU model in processing local information. By using Bayesian optimization algorithms to automatically adjust hyperparameters, we can better capture the time series characteristics and complex patterns of stock prices The experimental results show that the model outperforms the standalone BIGRU model, the ARIMA model, and the unoptimized ARIMA-BIGRU model in terms of prediction accuracy and training time, effectively handling short-term stock price forecasting issues.