Analysis Of Stock Forecasting Based On Attention Mechanism

In the application field where statistics and finance intersect,the stock market is a non-linear system with intricate quantities and relationships in terms of variables.How to accurately predict its future price changes is a worthy research direction.Traditional statistical learning methods are difficult to handle such intricate relationships in practical applications,and it is difficult to accurately extract the informa contained in stock data.With the development of computer hardware and deep learning,more and more deep learning methods are used in the stock market prediction.The research content of this article mainly lies in how to use deep learning techniques to obtain more accurate results in the prediction of the stock market.In recent years,the attention mechanism has achieved great success in the field of natural language processing,and has also led to many applications in time series prediction.Because stock data is filled with a lot of noise,and the relationship between variables is also very complex,based on this nature,this paper proposes a model based on the attention mechanism to predict stock prices.The encoder-decoder structure uses long short term memory neural networks and attention mechanisms to extract information in the historical data of the stock market,and a two-stage attention mechanism is added to the encoder section,which can be better from the horizontal and vertical directions,respectively.The relationship between the hidden states of the data can be captured more accurately,and the internal information in the historical data of the stock can be more accurately extracted,and an attention mechanism is added to the decoder to predict the future trend of stock fluctuations.Experimental comparison,the accuracy of the improved model is further improved.Through the analysis of the experimental results,we can see that the accuracy rate of the multi-stage stock trend prediction model is significantly improved compared with other classic stock trend prediction models.