Public Sentiment Analysis in Twitter Data for Prediction of a Company's Stock Price Movements
Authors:
- Li BingLBLi Bing
- Keith C.C. ChanKCKeith C.C. Chan
- Carol OuCOCarol Ou
- 1 November 2014
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 232-239
Abstract
There has recently been some effort to mine social media for public sentiment analysis. Studies have suggested that public emotions shown through Tweeter may well be correlated with the Dow Jones Industrial Average. However, can public sentiment be analyzed to predict the movements of the stock price of a particular company? If so, is it possible for the stock price of one company to be more predictable than that of another company? Is there a particular kind of companies whose stock price are more predictable based on analyzing public sentiments as reflected in Twitter data? In this article, we propose a method to mine Twitter data for answers to these questions. Specifically, we propose to use a data mining algorithm to determine if the price of a selection of 30 companies listed in NASDAQ and the New York Stock Exchange can actually be predicted by the given 15 million records of tweets (i.e., Twitter messages). We do so by extracting ambiguous textual tweet data through NLP techniques to define public sentiment, then make use of a data mining technique to discover patterns between public sentiment and real stock price movements. With the proposed algorithm, we manage to discover that it is possible for the stock closing price of some companies to be predicted with an average accuracy as high as 76.12%. In this paper, we describe the data mining algorithm that we use and discuss the key findings in relation to the questions posed.Keywords
This publication has 10 references indexed in Scilit:
- Twitter mood predicts the stock marketJournal of Computational Science, 2011
- Predicting the Future with Social Media2010
- Twitter power: Tweets as electronic word of mouthJournal of the American Society for Information Science and Technology, 2009
- Textual analysis of stock market prediction using breaking financial newsACM Transactions on Information Systems, 2009
- The dynamics of viral marketing2006
- Predicting box-office success of motion pictures with neural networksExpert Systems with Applications, 2006
- The predictive power of online chatter2005
- The Real Power of Artificial MarketsScience, 2001
- Forecasting stock indices: a comparison of classification and level estimation modelsInternational Journal of Forecasting, 2000
- Learning sequential patterns for probabilistic inductive predictionIEEE Transactions on Systems, Man, and Cybernetics, 1994