Prediction Of Shanghai-shenzhen 300 Index Based On XGBoost-LSTM Neural Network

The emergence and development of China’s stock market up to now,after 30 years of continuous development and reform,has gradually become one of the main forces to promote the growth of our national economy,is an indispensable part of our financial market.It is precisely because the stock market of our country occupies an important position in the domestic financial market that the fluctuation trend of stock price is influenced by many influencing factors,such as domestic and foreign economic policy and international trend,and all kinds of influencing factors restrict and interfere with each other,which makes the fluctuation trend of stock price of our country stock market have random characteristics.Stock data,as the traditional financial time series data,has the characteristics of dynamic nonlinear,highly noisy and non-parametric data.Therefore,it is very important to construct a reasonable forecast model of stock market data and make a reasonable forecast of stock market data.the xgboost-lstm neural network model designed in this paper innovatively combines the gradient-lifting decision tree(xgboost)model with the short-and-shortterm memory neural network(lstm)to predict stock market data.the gradient-lift decision tree(xgboost)model is the current hot machine learning model,which not only performs well in data prediction,but also performs better in feature selection.in this paper,we use the gradient lifting decision tree(xgboost)to select the more important features from the many characteristic indicators that affect the stock price trend and fluctuation: volunm and ma5,which are composed of feature data set with closing price,opening price,highest price and lowest price,as the input data of the short-and-short-term memory neural network(lstm)model.Finally,a prediction model of the CSI 300 index based on the XGBoost-LSTM neural network was established,a control group and a control model were set up,and various measurement indexes were established.The experimental results were analyzed from two perspectives,namely,intuitive perspective and measurement index.The experimental results show that the XGBoost-LSTM neural network model designed in this paper has better performance in stock price prediction.By comparing and analyzing the results of the XGBoost-LSTM neural network model designed in this paper and the results of the modified long-term and short-term memory neural network(LSTM),the prediction performance of the gradient-enhanced decision tree(XGBoost)model for the long-term and short-term memory neural network(LSTM)model is demonstrated.There is a big improvement.Through the comparative analysis of the experimental results of the modified long-term and short-term memory neural network(LSTM),artificial neural network(ANN),and moving average autoregressive(ARIMA)model,the improved long-term and short-term memory neural network is not only from an intuitive perspective but also from a measurement index perspective.Compared with other prediction models,the prediction results of(LSTM)are better,which indicates that the XGBoost-LSTM neural network designed in this paper is more suitable for predicting the fluctuations and trends of stock prices.However,the performance of the prediction accuracy of the direction of the stock price index is not outstanding.Therefore,the XGBoost-LSTM neural network model designed in this paper should be further optimized and improved in the prediction accuracy of the direction of the stock price index.