Prasetya, Agung Setiawan (2017) Model Gstar Dengan Variabel Eksogen Metrik Dan Non Metrik Untuk Peramalan Inflasi Di Kalimantan. Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
Tesis 1315201708.pdf - Published Version Download (7MB) | Preview |
Abstract
Salah satu indikator ekonomi makro yang digunakan dalam penyusunan kebijakan pemerintah di bidang ekonomi adalah inflasi. Inflasi merupakan data time series bulanan yang diduga dipengaruhi oleh aspek antar lokasi. Salah satu metode time series multivariat yang menggabungkan unsur dependensi waktu dan lokasi (space time) adalah model Generalized Space Time Autoregressive (GSTAR). Dalam perkembangan model space time, tidak hanya dipengaruhi oleh dependensi waktu dan lokasi, tetapi juga terdapat faktor lain yang bisa digunakan untuk menambah akurasi dalam peramalan yaitu berupa variabel eksogen. Model
GSTAR dengan melibatkan variabel eksogen dikenal dengan model GSTARX. Variabel eksogen yang digunakan adalah skala metrik (curah hujan) dan skala non metrik yaitu variasi kalender dan intevensi berupa kenaikan harga bahan bakar minyak (BBM). Studi kasus dalam penelitian ini diterapkan untuk peramalan inflasi enam kota di Kalimantan yaitu Pontianak, Sampit, Palangkaraya,
Banjarmasin, Balikpapan dan Samarinda. Tujuan penelitian ini adalah ingin mendapatkan model GSTARX yang sesuai untuk peramalan inflasi enam kota di Kalimantan, sehingga hasil ramalannya bisa dijadikan informasi awal bagi
pemerintah dalam menentukan kebijakan. Hasil pemodelan GSTARX untuk inflasi enam kota di Kalimantan adalah GSTARX-GLS ([1,12]). Pemodelan univariat dengan menambahkan variabel eksogen memberikan nilai RMSE yang
lebih kecil dibandingkan tanpa melibatkan variabel eksogen. Demikian juga tingkat akurasi peramalan menunjukkan bahwa model univariat lebih baik
dibandingkan dengan GSTARX-GLS. Hal ini berdasarkan dari nilai RMSE outsample yang minimum.
==================================================================================================================
One of the macroeconomic indicators that used in the formulation of government’s economic policy is inflation. Inflation is a monthly time series that also is influenced by location effects. Generalized Space-Time Autoregressive (GSTAR) is a multivariate time series model that combines time and location effects. The space-time data is not only influenced by time and inter-dependencies of location, but also there are other factors to increase the accuracy of forecasting time series, that can be expressed in exogenous variables. GSTAR model involving exogenous variable is known GSTARX model. The exogenous variable consists of the metric and non-metric scales. In this research, exogenous variables were taken into consideration as metric scale i.e. rainfall and non-metric scales that are the calendar variation and intervention in the form of the increase of fuel price. The case study is applied of GSTARX for forecasting inflation in six cities in Kalimantan i.e. Pontianak, Sampit, Palangkaraya, Banjarmasin, Balikpapan and Samarinda. The objection of this research is to obtain appropriate GSTARX model for inflation prediction so that the results of prediction can be used early information for government decision of policy. GSTAR modeling results for the inflation in six cities in Kalimantan is GSTAR([1,12]1). By using the inverse distance weighting, showed that inflation in a region influenced by other regions in the previous twelve month period. The empirical result of GSTARX model for the inflation in six cities in Kalimantan is GSTARX-GLS ([1,12]1). Modelling by univariate gives better results than model GSTARX-GLS. It is shown by the smallest of RMSE at in-sample dataset. Similarly, the accuracy of forecasting using out-sample RMSE shows that the univariate model is better than GSTARX-GLS.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | GSTARX, Inflation, Kalimantan, space-time, time series |
| Subjects: | H Social Sciences > HB Economic Theory H Social Sciences > HB Economic Theory > Economic forecasting--Mathematical models. H Social Sciences > HC Economic History and Conditions |
| Divisions: | Faculty of Mathematics and Science > Statistics > 49101-(S2) Master Thesis |
| Depositing User: | AGUNG SETIAWAN PRASETYA |
| Date Deposited: | 19 Apr 2017 04:30 |
| Last Modified: | 06 Mar 2019 08:41 |
| URI: | http://repository.its.ac.id/id/eprint/3273 |
Actions (login required)
 |
View Item |