WorldCIST'15 - 3rd World Conference on Information Systems and Technologies

Full Program »

Big Data for Stock Market by means of Mining techniques

Predict and react to future events are the major advantages to any company. Big Data comes up with huge power, not only by the ability of processes large amounts and variety of data at high velocity, but also by the capability to create value to organizations. This article presents an approach to a Big Data based decision making in the stock market context. The correlation between news articles and stock variations it is already proved but it can enrich with other indicators. In the project developed where collected news articles from three different web sites and the stock history from the New York Stock Exchange. In order to proceed to data mining classification algorithms the articles were labeled by their sentiment/impact, the direct relation to a specific company and geographic market influence. With the proposed model it is possible identify the patterns between this indicators and predict stock price variations with accuracies of 100 percent. Moreover the model shown that the stock market is very sensitive to news about generic topics like government society and environment and the impact it will also depend of the company geographic cover.

Author(s):

Luciana Lima    
Algoritmi Research Centre, University of Minho
Portugal

Filipe Portela    
Algoritmi Research Centre, University of Minho
Portugal

Manuel Santos    
Algoritmi Research Centre, University of Minho
Portugal

José Machado    
Algoritmi Research Centre, University of Minho
Portugal

António Abelha    
Algoritmi Research Centre, University of Minho
Portugal

 

Powered by OpenConf®
Copyright ©2002-2013 Zakon Group LLC