Improved BIGRU Model and Its Application in Stock Price Forecasting

In order to obtain better prediction results, this paper combines improved complete ensemble EMD (ICEEMDAN) and the whale algorithm of multi-objective optimization (MOWOA) improve bidirectional gated recurrent unit (BIGRU), which makes full use original complex stock price time series data improves hyperparameters BIGRU network. To address problem that cannot make stationary data, sequence are processed using ICEEMDAN decomposition derive non-stationary parts modeled with autoregressive integrated moving average model (ARIMA), respectively. The modeling process introduces a for optimization, probability finding best combination parameter vectors. R2, MAPE, MSE, MAE, RMSE values algorithm, ICEEMDAN-BIGRU MOWOA-BIGRU were compared. An improvement 14.4% over algorithm’s goodness-of-fit value will greatly accuracy predictions.