Using Genetic Algorithm Optimization SVM to Construction of Investment Model

At the mo ment, S upport Ve ctor Machine ( SVM) has been widely u sed i n t he study of stock investment related topics. Stock investment can be further divided into three s trategies such as: buy, sell and hold. Using data concerning China Steel Corporation, this article adopts genetic algorithm for the search of the best SVM parameter and the selection of the best SVM prediction variable, then it will be compared with Logistic Regression for the classification prediction capability of stock investment. From the classification prediction result and the result of AUC of the models presented in this article, it can be seen that the SVM after adjustment of input variables and parameters will have classification prediction capability relatively superior to that of the other three models.