Extracting the best features for predicting stock prices using machine learning

摘要

Predicting stock price is always a challenging task-In this paper we are trying to predict the next day's highest price for eight different companies individually. For this we are using different feature sets to predict the price. It is observed that the Volume+Company and Nasdaq+S & P 500 +Company sets performed better than any other feature sets used. Also these features were very helpful for predicting stock price using sequential minimal optimization (SMO) and bagging approach. Comparing different methods, the best results were obtained using SMO and bagging.