1. Türkiye’de Enflasyon Oranlarının Makine Öğrenme Yöntemi ile Tahmini.
- Author
-
NAS, Serkan, AKBOZ CANER, Ayşe, and ERGİN ÜNAL, Ayşe
- Subjects
- *
INTEREST rates , *ARTIFICIAL neural networks , *RANDOM forest algorithms , *MONEY supply , *DECISION trees - Abstract
One of the important factors of decision units and policymakers to develop successful policies is the correct estimation of macroeconomic variables for future periods. Inflation is among these macroeconomic indicators, and to realize successful policies, it is necessary to minimize the real effects and severity of inflation, determine future changes and their effects, and predict inflation reliably. Accurate forecasting of inflation is important both in terms of the policies to be implemented and the investment decisions to be taken by the public and private sectors. In this context, quarterly data were selected for Turkey using the 2008-2023 time period. In the study, machine learning methods were used instead of traditional econometric methods, in which the difference between prediction and reality was minimized. Brent oil, Rediscount advance interest rate, money supply, CPI, tax revenues, general budget revenues, policy rate, US Dollar/TL parity, and GDP attributes, which are thought to affect inflation, were selected by using random forest, decision tree and multi-layered detector, which is an artificial neural networks method, which are alternative machine learning methods. With the selected attributes, it is aimed to determine the method and influencer to make the correct estimation. The analysis results show that the decision tree model predicts the most accurate inflation rate compared to the random forest and multilayer sensor. Another finding from the study is that the highest determinant of inflation in Turkey is the US dollar. [ABSTRACT FROM AUTHOR]
- Published
- 2024
- Full Text
- View/download PDF