A PCA-IGRU Model for Stock Price Prediction

<p>Accurate stock price prediction is significant for investors to avoid risks and improve the return on investment. Stock a typical nonlinear time-series problem, which many factors affect. Still, too much analysis of influencing will lead input redundancy large amount computation in model. Although model based Recurrent Neural Network (RNN) has good effect, it problem oversaturation. This paper proposes closing Principal Component Analysis (PCA) Improved Gated Unit (IGRU), PCA-IGRU. PCA can reduce information without destroying correlation original data, thus reducing time training prediction. IGRU an improved (GRU) model, prevents oversaturation by introducing Anti-oversaturation Conversion Module (ACM) enhances sensitivity learning. selects trading data Shanghai Composite Index (SCI) China as experimental data. The PCA-IGRU compared with seven baseline models. results show that better accuracy shorter time.</p> <p> </p>