A Novel Approach to Evaluate Rice Prices using Bayesian GSTARI (1,1,1) Model

Spatial data refers to the information regarding the price of rice obtained from several regions in East Java, namely Kediri, Blitar, Malang, Probolinggo, Pasuruan, Mojokerto, Madiun, Surabaya and Batu. The data pertainsto rice prices collected throughout the period from January 2019 to July 2023 can be classified as a time series. Thus the rice price data for nine East Java regions from January 2019 to July 2023 are spatiotemporal. This data can be analyzed using the statistical model Generalized Space Time Autoregressive Integrated (GSTARI). This model is a modification of GSTAR designed for non-stationary data. Numerous previous studies have been conducted regarding the spatial modeling of rice prices. However, the use of Bayesian estimation methodology within a GSTARI model remains unexplored. In estimating model parameters, the Bayesian approach has advantages over the non-Bayesian approach; namely, the small sample size and free distribution of the data. In this paper, we construct the GSTARI (1, 1, 1) model with a Bayesian estimator in order to model East Java’s rice prices. The weight used in this investigation is Inverse Distance. According to the findings of this investigation, all GSTARI (1, 1, 1) parameters estimated using the Bayesian estimator are significant at all locations. With the OLS estimator, the GSTARI model parameters (1, 1, 1) were not significant at all locations. The GSTARI (1, 1, 1) model with Bayesian estimator also yields RMSE and AIC values that are much smaller than the OLS estimator.