BAREKENG: Jurnal Ilmu Matematika dan Terapan
Vol 19 No 1 (2025): BAREKENG: Journal of Mathematics and Its Application

APPLICATION OF THE GENERALIZED SPACE TIME AUTOREGRESSIVE (GSTAR) METHOD IN FORECASTING THE CONSUMER PRICE INDEX IN FIVE CITIES OF SOUTH SULAWESI PROVINCE

Zaki, Ahmad (Unknown)
Shafruddin, Lutfiah (Unknown)
Thaha, Irwan (Unknown)



Article Info

Publish Date
13 Jan 2025

Abstract

Changes in the Consumer Price Index (CPI) over time reflect the rate of increase (inflation) or decrease (deflation) of goods and services for daily household needs. The CPI and inflation serve as barometers for economic growth stability, as controlled inflation can increase people's purchasing power over time. According to the Central Statistics Agency (2023), in December, the year-on-year (y-o-y) inflation for five cities in South Sulawesi (Bulukumba, Watampone, Makassar, Parepare, and Palopo) was 2.81 percent, with a CPI of 117.35. Of the five cities, the highest y-o-y inflation occurred in Makassar at 2.89 percent, with a CPI of 117.49, while the lowest y-o-y inflation occurred in Palopo at 2.21 percent, with a CPI of 115.60. CPI forecasting is one way to predict future inflation values. This study aims to develop the best GSTAR model for forecasting CPI data for five cities in South Sulawesi, a topic that has not been extensively covered in previous research. The goal is to provide valuable information for maintaining CPI stability in South Sulawesi and to support the formulation of better economic policies. The study focuses on five cities within South Sulawesi, where direct relationships between cities are possible, allowing the spatial model to be limited to the first-order. The data used in this study consists of monthly CPI data from January 2014 to March 2023. The location weights used in the model include uniform weights, inverse distances, and normalized cross-correlations. The model development steps include testing for data stationarity, determining the space-time sequence, calculating location weights, estimating parameters, testing model adequacy, comparing Root Mean Square Error (RMSE), and selecting the best model for forecasting. The best GSTAR model found is GSTAR (1;1)-I(2) with inverse distance weighting, which yielded the smallest RMSE value. The results show that the forecasted values closely match the actual values for each city from March to September 2023.

Copyrights © 2025






Journal Info

Abbrev

barekeng

Publisher

Subject

Computer Science & IT Control & Systems Engineering Economics, Econometrics & Finance Energy Engineering Mathematics Mechanical Engineering Physics Transportation

Description

BAREKENG: Jurnal ilmu Matematika dan Terapan is one of the scientific publication media, which publish the article related to the result of research or study in the field of Pure Mathematics and Applied Mathematics. Focus and scope of BAREKENG: Jurnal ilmu Matematika dan Terapan, as follows: - Pure ...