Association Rules Analysis with Apriori Algorithm and Data Visualization: Evidence from Borsa Istanbul
Öz
Considering that association analysis can be applied in finance, as it is in many fields, a study was conducted using stock prices. This study utilized two datasets covering the stocks included in the BIST TUM index and the BIST TUM-100 index, which were calculated within Borsa Istanbul. The data were analyzed using the Apriori algorithm and supported by data visualization techniques. The research used daily closing data of 421 companies in the BIST TUM index and 340 companies in the BIST TUM-100 index, obtained from the Finnet database. The data belongs to the period 10.01.2022/08.01.2024, and the association analysis was performed with stock closing prices according to the interestingness rule criteria. ARules () and ARulesViz () packages were used for the Apriori algorithm in the analysis performed in the R programming language, and Shiny () was used for the representation of codes and visuals. In Model 1, which is based on BIST TUM index stocks, 66 association rules were obtained, and in Model 2, which covers BIST TUM-100 index stocks, 34 association rules were obtained. The study concludes that all the parameters analyzed reveal significant and interesting rules, as statistically evidenced by Fisher’s exact p-values, indicating that the companies may be significantly related.
Anahtar Kelimeler
Portfolio Management, Data Mining, Apriori Algorithm, Fisher’s Exact Test, Interestingness Measures, Stocks
Proje Numarası
1129B372300796
Etik Beyan
Acknowledgments: The second author of the article has utilized the knowledge gained from the “TÜBİTAK 2237-A BİDEB Statistical Modeling Techniques and Applications in Natural Sciences (Project No:1129B372300796)” and “TÜBİTAK 2237-A BİDEB Natural Science-Based Exploratory Data Analysis and Data Visualisation (Project No:1129B372400933)” projects, supported by TÜBİTAK, during the article process.
Kaynakça
- Ataman, G., & Kahraman, S., 2022. Comparing decision trees and association rules for stock market expectations in BIST100 and BIST30. Scientific Annals of Economics and Business, 69(3), 459-475. https://doi.org/10.47743/saeb-2022-0024.
- Aydemir, E., 2019. Weka ile yapay zeka makine öğrenmesi, Yapay sinir ağları-derin öğrenme, Ankara.
- Aydemir, E., & Yavuz, M., 2019. Mevsimlere göre ilaç satış verilerinin birliktelik analizi ile incelenmesi. Uluslararası Yönetim Bilişim Sistemleri ve Bilgisayar Bilimleri Dergisi, 3(1), 23-30. https://doi.org/10.33461/uybisbbd.582061