Predicting Behaviors of Stock Market
Adhvik Shetty1, Subham Chatterjee2, Parimala R3
1Adhvik Shetty, PES Institute of Technology, BSK III Stage, angalore,, Karnataka, India
2Subham Chatterjee, PES Institute of Technology, BSK III Stage, angalore, Karnataka, India.
3Professor Parimala R, Professor, Information Science Department, ES Institute of Technology, BSK III Stage, Bangalore, Karnataka, India.
Manuscript received on June 18, 2015. | Revised Manuscript received on June 26, 2015. | Manuscript published on July 05, 2015. | PP: 120-123 | Volume-5 Issue-3, July 2015. | Retrieval Number: C2676075315/2015©BEIESP
Open Access | Ethics and Policies | Cite
©The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/
Abstract: Prices of stock depend on a variety of factors. Predicting and building a model is a daunting task to any analyst. To predict the behavior of stock market, one goes through the company news, economic and political news and global sentiments. Considering the large number of news articles, there are some which can be missed out. Also it is impossible to focus on each and every news article as soon as it is published on the internet. In this paper, we analyze the sentiment generated by news articles and correlate the sentiment with the actual change in stock market prices. This gives a deeper insight into the correlation and tells us how much news articles influence the stock market. After extensive research we have decided to use a hybrid technique involving machine learning and natural language processing concepts. We have used n–gram as the feature creation, chi square as the feature selection and support vector machines as the classification technique. Improving the accuracy of predicting stock market trends, we hope to aid investors in better decision making based on real time sentiment of news articles.
Keywords: PRICES, CLASSIFICATION, ARTICLES, NEWS, POLITICAL.