Sensitivitas Ukuran Amatan Model Autoregresi Fuzzy

Abstract

The conventional time series modelling method is Autoregressive Integrated Moving Average (ARIMA). ARIMA method is more effective if it is applied in long time periode of observations that at least 50 observations. Tseng on 2001, introduced Fuzzy ARIMA method to deal time series modelling on short time periode of observations. Fuzzy ARIMA combines the advantages of fuzzy logic, ARIMA and fuzzy regression. Considering Tseng’s study the aim of this study is tracing the optimal observation range size when fuzzy ARIMA method is applied at the curent condition. The result of this study indicated that fuzzy ARIMA when the condition of time series data AR(2) with μ=50 and parameter estimation 1=0.600 and 2=0.300 was more sensitive when the observation size less than 20.