Veri Madenciliği Büyük Veri Hibrit Makine Öğrenmesi Regresyon
For investors, the prediction of the trend of the stock market indices is not only a difficult but needed issue due to various variables that affect the index. Global events, policymakers' actions, economic variables and, in particular, recent epidemics have caused indices to respond more than once. Using hybrid regression models, the study aims to predict the relationship between BİST 100 index and G20 index. The closing values of 2,432 days are taken as the data set by considering the common days when the index days between 2010/01/01 and 2019/12/31 are open. Independent variables, Bovespa, Italy40, KOSPI, Nikkei 225, BMVIIPC, Shanghai and Tadawul, are obtained by taking into account the correlation values of indices which are expected to have an effect on the BİST index in the G20 countries, with an absolute value below 0.80. In the study, data mining applications have found the highest R2 = 0.9918 for 30 segmentation with the Knime program, the lowest MAE = 0.0650, MSE = 0.0082 and RMSE = 0.0903. According to the results of the analysis, it is determined that Bovespa, Italy40, KOSPI, Nikkei 225 and BMVIIPC indices affect BIST-100 index positively, and Shanghai index negatively.
Birincil Dil | Türkçe |
---|---|
Konular | İşletme |
Bölüm | Araştırma Makaleleri |
Yazarlar | |
Yayımlanma Tarihi | 20 Ağustos 2021 |
Gönderilme Tarihi | 27 Kasım 2020 |
Yayımlandığı Sayı | Yıl 2021 Cilt: 12 Sayı: 31 |