PEMODELAN HARGA MINYAK DENGAN SMOOTH TRANSITION AUTOREGRESSIVE

Abstract

Observation data series of oil price are mutually dependent and fluctuating. In modeling the world oil price data that exist in Indonesia needed the method of nonlinear time series, because the data showed non-linear pattern. Method used in the writing of this thesis is the Smooth Transition Autoregressive (STAR) which is an extension of the autoregressive form. In the STAR modeling of variables determined transition (st) of the transition function G (st, Y, c). The author defines st become zt_k, where k is the order of p in ARIMA model (p, d, q) which pass through nonlinearity test for STAR models. Transition function G (st, Y, c) was also obtained from the nonlinear model of the STAR test results. From nonlinearity test results, obtained the transition variable zt_3 and logistics functions of first-order transition. Pass through discussion and analysis of the results obtained from the STAR model of oil prices that went to Indonesia is (Azt_3; 0.42772; -1.36607).