Research On Stock Prediction Based On Support Vector Machine

With the development of China's economic construction and the improvement and maturity of the financial market,more and more investors choose to buy stocks as their own investment methods.How to grasp the price trend of stocks,conduct effective stock investment management and improve stock investment efficiency has become the most concerned problem to investors.In this paper,two kinds of stock prediction models are built based on support vector machine to predict the future stock price and trend respectively.This article firstly uses the empirical mode decomposition method to decompose the stock price sequence to obtain some more stable intrinsic mode functions and the residual than the original series.Then,a hybrid kernel function is constructed,and the grasshopper optimization algorithm is used to optimize the parameters of the SVR.Support vector machine regression is used to analyze and predict the intrinsic mode functions and the residual after decomposition.Through empirical analysis and model comparison,it can be seen that the EMD-GOA-SVR model constructed has better prediction accuracy.This paper also considers the differences in the impact of different features and different sample points on the model prediction,and combines the weightedsupport vector machine with the Relief algorithm.while considering the sample distance weighting,the Relief algorithm is used to calculate the influence degree of each feature in the classification,that is,the weight of each feature.Then,the weighted features are input to the weighted support vector machine for training to predict the future trend of the stock price.The numerical experiment results of six stocks in different industries show that the prediction accuracy of the constructed Relief-WSVM model for stock price trend is more than 70%,which can provide valuable reference for investors.