Γραμμική παλινδρόμηση για την πρόβλεψη της ανόδου ή πτώσης του χρηματιστηρίου
View/
Keywords
Μηχανική μάθηση ; Γραμμική παλινδρόμηση ; Πρόβλεψη χρηματιστηρίουAbstract
The scope of this thesis was the development and the evaluation of linear regression for the prediction of the international stock market MSCI rise/decline. Particularly, we assumed that there is a linear correlation between the MSCI index and the most important indices around the world along with the polarity of the news (i.e. if the news are positives or negatives). The indices we used are, the stock market of Frankfort (DAX), the stock market of Paris (CAC 40), the Dow Jones, the stock market of London (FTSE 100), the NASDAQ and the S&P 500.
The findings of the experiments, conducted in the context of the thesis, are interesting both for the various stock market indices studied as well as for the contribution of additional information from the collected texts.
Finally, the thesis is accompanied by a graphical user interface, which enables people who are interested to this topic to train their own linear regression models, loading training sets. People can also predict the stock market rise or decline by giving real time prices to the independent variables.