MODELING DATA FUZZY TIME SERIES WITH USING THE DECOMPOSITION OF SINGULAR AND VALUE THE APPLICATION OF INFLATION INFLUENCE LEVEL IN INDONESIA

Agus Maman Abadi (FMIPA UNY)

Jurnal Penelitian Saintek
Vol 14, No 1: April 2009

The aims of this research are to construct a new method for modeling fuzzy time series data and to apply the method for forecasting Indonesian inflation rate. The procedure of this research is done by the following steps: (1) determine fuzzy relations using table lookup scheme, (2) Apply the singular value decomposition to reduce the unimportant fuzzy relations, (3) apply  the method to forecasting Indonesian inflation rate. The result of this research is that it was designed a new method to construct the fuzzy time series model using singular value decomposition method. Then, the method is applied to forecast the Indonesian inflation rate based on fuzzy time series data. Forecasting inflation rate using the proposed method yields a higher accuracy than that using table lookup scheme and neural network methods.