Stock Prediction Methods based on Ensemble Learning

Abstract

With the rapid development of stock market, there have been large interests in stock prediction. The decision making based on rational and logical analysis as well as forecast often has a very positive supporting effect, reducing investment risk while enhancing the profits. The development of technology has led to a variety of mature machine learning models for predicting the stock market such as the support vector machine (SVM) model and support vector regression (SVR) model, which will be introduced later in the paper. In this paper, it focuses on the improvement of the existing machine learning models by comparing the deviation and coefficient of curves of different stocks. The experiment indicates that the ensemble models provide more effective and more accurate stock prediction compared with only using the SVR model.