Abstract:

The purpose of this study was to discover how to improve accuracy of natural gas consumption forecasts of different time horizon and answer how this knowledge can be used to advance theory and practice of natural gas time series forecasting. We analyzed and compared classic approach containing ARIMA, ETF, THETA and ANN models. Than we used fuzzy approach to quantify uncertainty and compare this attitude with classic crisp forecasting. The analysis was based on dataset which contains time series from biggest European countries. We found out that fuzzy approach gave better forecasts mainly in short-term forecasting in comparison to other methods.