Setyowati, Endah (2020) Model Hybrid Generalized Space-Time Autoregressive – Elman Recurrent Neural Network untuk Peramalan Data Space-Time dengan Variabel Eksogen. Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
06211850010012-Master_Thesis.pdf Download (10MB) | Preview |
Abstract
Generalized Space-Time Autoregressive (GSTAR) merupakan model peramalan data spatio-temporal. Selanjutnya dikembangkan model GSTARX yang bertujuan untuk menangkap komponen deret waktu, seperti tren, musiman, dan variasi kalender dari data spatio-temporal. Selanjutnya model peramalan GSTARX dapat digabung dengan model peramalan nonlinier, salah satunya Elman RNN. Sehingga pada penelitian ini diusulkan model hybrid Generalized Space-Time Autoregressive dengan variabel eksogen dan Elman Recurrent Neural Network atau GSTARX-Elman RNN. Penelitian ini terdiri dari dua kajian, yaitu kajian simulasi dan kajian terapan. Kajian simulasi menggunakan data bangkitan yang terdiri dari tren, musiman, dan variasi kalender, dengan menggunakan dua skenario residual, yaitu residual yang mengikuti model linier dan nonlinier. Hasil yang diperoleh dari kajian simulasi ini menunjukkan bahwa model GSTARX-Elman RNN mampu memodelkan dengan baik data simulasi yang mengandung residual yang mengikuti model linier maupun nonlinier. Sedangkan kajian terapan mengaplikasikan data inflow dan outflow total dan pecahan besar pada empat Kantor Bank Indonesia (KBI) di Jawa Tengah, yaitu Semarang, Solo, Purwokerto, dan Tegal. Hasil kajian terapan menunjukkan bahwa model GSTARX-Elman RNN mampu memodelkan dengan baik data inflow dan outflow, terutama pada inflow total, Rp 100.000,00, Rp 50.000,00 dan outflow Rp 50.000,00. Hasil tersebut sesuai dengan hasil dari M4-Competition yang menyebutkan bahwa model peramalan hybrid cenderung memberikan hasil peramalan yang lebih akurat daripada model peramalan individu
===============================================================================================================================
Generalized Space-Time Autoregressive (GSTAR) is a spatio-temporal data forecasting method. The GSTARX model aims to capture time-series components, such as trends, seasonal, and calendar variations from spatio-temporal data. The GSTARX method can be combined with nonlinear methods, i.e., Elman RNN. Hence, in this research proposes a hybrid method by combining Generalized Space-Time Autoregressive with exogenous variables and Elman Recurrent Neural Network (GSTARX-Elman RNN). This research consists of two studies, namely simulation and applied studies. The simulation study uses generating data consisting of trends, seasonal, and calendar variations, using two error scenarios, i.e., linear and nonlinear error. The result of these simulations showed that the GSTARX-Elman RNN model can model well the simulation data that contains linear and nonlinear error. Whereas the applied study uses total and large fractions of inflow and in four Bank Indonesia Offices in Central Java, i.e., Semarang, Solo, Purwokerto, and Tegal. The results showed that the GSTARX-Elman RNN model can model well the inflow and outflow data, especially the total inflow, IDR 100,000.00, IDR 50,000.00, and outflow IDR 50,000.00. These results are consistent with the M4-Competition result that the hybrid models tend to provide more accurate forecast performance than individual forecast models.
| Item Type: | Thesis (Masters) |
|---|---|
| Additional Information: | RTSt 519.535 Set m-1 2020 |
| Uncontrolled Keywords: | Hybrid GSTARX-Elman RNN, Inflow, Jawa Tengah, Outflow, Space-Time |
| Subjects: | Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA280 Box-Jenkins forecasting Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) Q Science > QA Mathematics > QA76.9 Computer algorithms. Virtual Reality. Computer simulation. |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis |
| Depositing User: | Endah Setyowati |
| Date Deposited: | 08 May 2023 01:40 |
| Last Modified: | 08 May 2023 01:40 |
| URI: | http://repository.its.ac.id/id/eprint/73830 |
Actions (login required)
 |
View Item |