Twitter Emotional Profile Improves The Stock Market Values Forecast (Do emotions derived from Twitter have prediction power on the stock mark
Abstract: The vast expansion of online social platforms and increasing openness of their users created a great opportunity for researchers to obtain more detailed information about such complicated phenomena as public opinions and emotions. And since it was proven by behavioral economist long time ago that emotions take part in decision--making process, the researchers have initiated a new series of exciting studies that can be combined under the title "emotions derived from online social platforms predict the future" The author of this thesis joins this large group of researchers by investigating whether 7 emotions, namely Surprise, Happiness, Comfort, Calm, Frustration, Anger, and Sadness expressed by Twitter users from Singapore can influence SIMSCI closing values index over the time. The analysis was based on hourly values of all 8 variables obtained for the period from 12 September 2011 to 12 November 2011. With a help of Multilayer Perceptron network it was discovered that combination of all 7 emotions expressed within an hour T have predictive power on SIMSCI closing price of the hour T, the conclusion drawn from the fact that Relative Error has decreased by more than 40%. The historical values of each of the emotions taken separately or in combination with others didn't show any consistency in generating lower errors.
AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)