Investigation of the effects of economic news on BIST 100 Index by using data mining

In this master thesis we tried to investigate whether there is a relationship between changes on BIST 100 Indeks and domestic news published on internet. In this dimension we use the news published on Dunya newspaper's web site between the 01.04.2016- 01.10.2016 dates. Data sets which obtained by using news text are reviewed with data mining. Because of huge amount of data sets, we reduced them to small size. Feature vectors are converted to appropriate form to process in. Then these feature vectors classified in. Basic aim of classification process is to determine the belonging groups of the words. Words have three labels positive, neutral, and negative. Machine leraning algorithism have been applied to generated clusters. At the end of this application labels have been assigned to the words to determine their belonging clusters. By applying the Tf-Idf weighting methods to the words, weighting tables of the words formed. By looking at values and impacts on this table market situation compared. Key Words: BIST 100, Data Mining, Text Mining