Study On Stock PricePrediction Based On PSO And LSSVM

With the development of Chinese stock market,direct financing plays more and more important role in the financial market.The flow and configuration of social capital is closely related to the stock market.Capital could be obtained through financing in stock market.Raising funds and expanding production is inseparable from the stock market for enterprises.Investors can also invest in the stock market to obtain profits,thus it is more and more important to conduct the thorough research to the stock market With the development of deep learning,more and more researchers have applied it to various fields.Because of the complexity and nonlinearity of the stock market,and diversity of influential factors of stock market,traditional prediction models of stock price or yields have certain drawbacks.They are unable to effectively deal with the volatility of the stock market But deep learning can effectively find the implied relationship between data,better fitting with the nonlinear data,improving the accuracy of the prediction model.Therefore,it is of great significance for the stock market to introduce deep learning into the stock market prediction model to build a model with stronger generalization ability,improving the accuracy of prediction model of the stock market.In this literature,the parameters of phase space reconstruction are determined by c-c method,and then the parameters of least squares support vector machine are optimized by particle swarm optimization algorithm.Compared with other prediction methods,the results show that this method has higher prediction accuracy and smaller prediction error.This paper will improve on the basis of this literature to predict the closing price of stock.In other reference,parameters of phase space reconstruction and least-squares support vector machine are optimized separately.Parameters of phase space reconstruction are selected by C-C method,and parameters of least-squares support vector machine are optimized by particle swarm optimization.This method completely separates the phase space reconstruction from the least squares support vector machine and ignores the relationship between them.However,in the prediction process of chaotic time series,the parameter optimization process of phase space reconstruction and LSSVM model is interdependent,and the combination of the optimal level of each parameter cannot guarantee the optimal parameter combination.Therefore,in this paper,particle swarm optimization algorithm is used to simultaneously optimize the parameters of phase space reconstruction and least-squares support vector machine,and the combination of optimized phase space reconstruction and least-squares support vector machine is combined,so as to achieve the optimal combination of parameters and optimize the overall model.