Dana, I Made Gde Meranggi (2018) Model Hybrid GSTARX-ANN Untuk Peramalan Data Space-Time Dengan Efek Variasi Kalender (Studi Kasus : Data Inflow Dan Outflow Uang Kartal Di Bank Indonesia Wilayah Jawa Timur). Masters thesis, Institut Teknologi Sepuluh Nopember.
|
Text
1316201028-Master Thesis.pdf - Accepted Version Download (5MB) | Preview |
Abstract
Selain berdimensi waktu, data juga bisa berdimensi ruang yang dikenal dengan data space-time. Model space-time merupakan suatu model yang menggabungkan unsur dependensi waktu dan lokasi pada suatu data time series multivariat, salah satu model space-time adalah Generalized Space-Time Autoregressive (GSTAR). Model GSTAR memiliki keterbatasan yaitu tidak mampu memodelkan time series yang nonlinier, namun hal ini bisa diatasi dengan menerapkan model hybrid pada GSTAR. Pada penelitian ini akan dilakukan pemodelan hybrid GSTARX-ANN, dimana model GSTARX sebagai komponen linier yang melibatkan variabel prediktor, yaitu efek variasi kalender dan ANN sebagai komponen nonlinier. Model hybrid GSTARX-ANN merupakan model terbaik dalam meramalkan data simulasi yang mengandung komponen tren, musiman, variasi kalender, dan deret noise linier maupun nonlinier dibandingkan model VARX dan GSTARX. Data inflow dan outflow di KPw BI wilayah Jawa Timur dipengaruhi oleh tren kebijakan BI, musiman, dan variasi kalender, serta mengandung unsur nonlinearity. Pengaruh minggu terjadinya hari raya Idul Fitri juga berpengaruh terhadap tingginya inflow dan outflow. Pada pemodelan data inflow di KPw BI wilayah Jawa Timur, model GSTARX-FFNN(8,2,1) bobot invers jarak merupakan model terbaik, sedangkan pada pemodelan data outflow model GSTARX-FFNN(8,15,1) merupakan model terbaik. ================================================================================================================== Aside from time dimension, data can also dimension of space known as space-time data. The space-time model is a model that combines elements of time and location dependencies in a multivariate time-series data. One of the space-time models is Generalized Space -Time Autoregressive (GSTAR). The GSTAR model has its limitations of not being able for modeling a nonlinear time series, this can be overcome by applying hybrid model on GSTAR. In this research will be modeling hybrid GSTARX-ANN, where GSTARX model as a linear component involving predictor variable, that is an effect of calendar variation and ANN as a nonlinear component. The hybrid GSTARX-ANN model is the best model for predicting simulation data containing trend, seasonality, calendar variations, linear and nonlinear noise series compared to VARX and GSTARX models. Inflow and outflow data in Bank Indonesia East Java region are influenced by BI policy, seasonal trend, calendar variation, and contain nonlinearity elements. GSTARX-FFNN (8,2,1) is the best model for modeling inflow data in Bank Indonesia East Java region , while GSTARX-FFNN (8,15,1) is the best model for modeling outflow data.
| Item Type: | Thesis (Masters) |
|---|---|
| Additional Information: | RTSt 519.536 Dan m |
| Uncontrolled Keywords: | ANN; GSTARX; GSTARX-ANN; space-time; variasi kalender |
| Subjects: | H Social Sciences > HA Statistics H Social Sciences > HA Statistics > HA30.3 Time-series analysis |
| Divisions: | Faculty of Mathematics, Computation, and Data Science > Statistics > 49101-(S2) Master Thesis |
| Depositing User: | I MADE GDE MERANGGI DANA |
| Date Deposited: | 02 Feb 2018 08:41 |
| Last Modified: | 02 Jul 2020 07:59 |
| URI: | https://repository.its.ac.id/id/eprint/51094 |
Actions (login required)
 |
View Item |
