Design And Implementation Of Star Market Stock Price Prediction System Based On Deep Learning

With the vigorous development of China’s economy,the financial market has attracted more and more Attention.Chinese President Xi Jinping announced the establishment of The Science and Technology Innovation Board(STAR Market)on the stock exchange of Shanghai in November 2018,which mainly serves the technological innovation enterprises that meet the national strategy,break through the key core technologies and have high market recognition.Therefore,it is of great guiding significance for economic development,the formulation of economic policies and the development planning of high-tech industries to correctly analyze and predict the future fluctuation and change trend of STAR Market.Based on deep learning,this thesis designs and implements a stock price prediction system of STAR Market.The main research contents are as follows:First of all,this thesis gets the stock related data through the crawler and Tushare API,and builds the data set.Secondly,this thesis uses TextCNN,LSTM,Bi-LSTM and Bi-LSTM+Attention neural network models to analyze investor sentiment based on stock comment data and financial news sentiment based on stock news data.Thirdly,the four features of stock price data,index data,financial news emotion analysis results and investor emotion analysis results are fused together as input features to predict the future trend of stocks through LSTM,Bi-LSTM and Seq2Seq neural network models,and the importance of input features to the prediction results is analyzed.Finally,Python and java development language are used to complete the construction of the stock price prediction system by using the Pytorch and Spring Boot.In the system,investors can select the stocks on the scitech innovation board for prediction and view the prediction results.