Research on Stock Selection Model based on Stochastic Forest Algorithms

Abstract

Stocking market is a very complicate non-linear system. The traditional quantitative stock selection models are always based on data without selecting the varieties, thus suffering from unneglected limitations. The article points out a SVM model based on the random forest algorithm, and improves the accuracy of the model using random forest addressing with the varieties. At the same time, by comparing the effects of random forest algorithm and principle component analysis algorithm in dimensionality reduction, the experiment shows that the SVM model based on the random forest algorithm preforms better than the SVM model based on the principle component analysis algorithm in accuracy, thus processing high level of practical value.