Forecasting consumer price index with multiplicative neuron model artificial neural networks

The forecasting of inflation is one of the most important applications of forecasting methods. In this thesis, Turkey inflation time series is forecasted by using classical and machine learning methods. Especially, the performance of the multiplicative neuron model artificial neural networks for inflations forecasting is compared with classical and other machine learning methods. A new forecasting algorithm that is based on multiplicative neuron model artificial neural networks are proposed in this thesis. As a result of applications, it is concluded that the proposed forecasting algorithm produced the best forecasting results for inflation forecasting.