Türkiye'de Bazı Temel Gıda Fiyatları İçin Yapay Sinir Ağları ve Zaman Serisi Tahmin Modellerinin Karşılaştırmalı Analizi
Comparative Analyses of Forecasting Models of Artificial Neural Network and Time Series Analyses for Selected Main Food Prices in Turkey

Summary/Abstract: In this study three main food products wheat barley and paddy prices are taken from Turkish Statistical Institute database monthly between the 2000-2014. Firstly these data are analyzed and the structure of data has determined. These datasets include trend but there is not a regular seasonality. Then traditional time series applications which are appropriate for these datasets, trend analyses, Holts double exponential smoothing, and non-seasonal Box-Jenkins (ARIMA) models and artificial neural network models are implemented with the help of computer programs (Matlab Neural Network Toolbox). As a result mean square errors (MSE) of the models are compared between each other. For the wheat and barley datasets Holts double exponential smoothing, ARIMA and neural network models give closer results. For paddy datasets neural network model give the best result.

• Details
• Contents