Long Short-Term Memory (LSTM) Network Applications in Stock Price Prediction

In response to the issues of low accuracy and stability in stock market prediction models, this paper introduces a CNN-BiLSTM-Attention prediction model. It consists of three components: firstly, a Convolutional Neural Network (CNN) layer is used to extract local features from historical stock data. These features are then passed to a Bidirectional Long Short-Term Memory (BiLSTM) layer, which is capable of capturing long-term dependencies in time series data and providing a deeper contextual understanding. Most importantly, this model incorporates an Attention mechanism that dynamically focuses on features at different time points, weighting them based on their importance to more accurately capture crucial information related to stock trends. Finally, we demonstrate the superiority of the CNN-BiLSTM-Attention stock prediction model by comparing its predictions with those of various original models. In the improved dataset, the performance of the CNN-BiLSTM-Attention model proposed in this paper is the best, with an average prediction accuracy of 62%.