Forecasting models have wide applications in decision making. In the real world, rapid changes normally take place in different areas, specifically in financial markets. Collecting the required data is a main problem for forecasters in such unstable environments. Forecasting methods such as Auto Regressive Integrated Moving Average (ARIMA) models and also Artificial Neural Networks (ANNs) need large amounts of historical data. Although fuzzy forecasting models such as fuzzy regression are suitable metods when the data available is scant, their performance is not satisfactory at times. In this paper, a new Fuzzy Auto Regressive Integrated Moving Average (FARIMA) is presented. The proposed model can be run with less data, so it is more suitable than other models for cases where there are limited data available. The results obtained on exchange rate forecasting reveal the efficiency of the proposed model.

Keywords: Exchange rate, Auto Regressive Integrated Moving Average, Fuzzy regression, Time series forecasting, Combined forecast.

Full-Text [PDF 301 kb]   (1330 Downloads)