Abstract:

In today’s competitive conditions, it has become important for investors to make accurate predictions from data. The diversity of data because of developments in information and technology has led to the need for modern statistical techniques. With these techniques, identifying and estimating unknown hidden relationships in the data is increasing day by day. Data mining is used in many fields, as well as in the field of finance. The data set used in this study consists of 1.423 trading days published between January 2, 2018 and June 27, 2023. The Fp Growth Algorithm was used to determine the co-change between the BIST30 Index, BIST100 Index, Dollar Rate, Euro Rate, Gold, and Brent Oil variables published in the daily bulletins, through association rule analysis, which is a descriptive model in data mining. After conducting the association analysis, a total of 20 association rules were generated, and the top 10 association rules were selected. The study observed a significant association at a confidence criterion of 0.99 among OIL, GOLD, BIST30, and BIST100 variables and 0.98 among OIL, EURO, BIST30, and BIST100 variables. We also observed that all of the following groups show statistically significant association at the 0.97 confidence level: USD, EUR, BIST30, BIST100; USD, GOLD, BIST30, BIST100; EUR, BIST30, BIST100; and USD, BIST30, BIST100.

Keywords: