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.