- Nişantaşı Üniversitesi Sosyal Bilimler Dergisi
- Cilt: 11 Sayı: 2
- COMPARISON OF PREDICTION PERFORMANCES OF REGRESSION MODELS IN MACHINE LEARNING: AN APPLICATION ON TH...
COMPARISON OF PREDICTION PERFORMANCES OF REGRESSION MODELS IN MACHINE LEARNING: AN APPLICATION ON THE TURKISH MERCANTILE EXCHANGE WHEAT INDEX
Authors : Hasan Arda Burhan
Pages : 602-623
Doi:10.52122/nisantasisbd.1377642
View : 65 | Download : 113
Publication Date : 2023-12-31
Article Type : Research
Abstract :Agricultural commodity prices have significant impacts on economies by leading to changes and regulations in both fiscal and monetary policies. These also have effects on household economies and consumer purchasing power particularly in developing countries. Thereby, instability and variability in these prices constitute adverse effects on these economies. On the other hand, assets of the commodity markets become popular just as bonds and stocks. Because of this growing interest, needs for managing risks, stable prices and lowering transaction costs has led to establishment of the commodity exchanges. In this context, Turkey put the licensed warehousing system into operation by founding the Turkish Mercantile Exchange (TMEX) to operate trades of Electronic Warehouse Receipts (EWRs). In this study, a model including US Dollar-Turkish Lira exchange rate (USD/TRY), Brent crude-oil prices, overnight interest rate and a daily dataset for the 01/04/2021-20/02/2023 period were used to assess several machine learning regression methods in predicting the TMEX Wheat Index (TMXWHT). As verified by comparisons with actual values and considering performance evaluation criteria, all methods yielded successful outcomes, furthermore, tree-based methods revealed better overall performance.Keywords : Emtia Piyasaları, Elektronik Ürün Senedi (ELÜS), TÜRİB, Makine Öğrenmesi Regresyon