Paembonan, Maya (2016) Model Arimax, Radial Basis Function Network (RBFN), Dan Hybrid Arimax-RBFN Untuk Peramalan Inflow Dan Outflow Uang Kartal Di Provinsi Papua. Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
1314201042-Master_Thesis.pdf - Accepted Version Download (2MB) | Preview |
Abstract
Dalam banyak kasus peramalan deret waktu, hubungan antar kejadian pada
waktu yang berbeda seringkali tidak linier. Dalam kondisi seperti ini, pendekatan
metode deret waktu linier dianggap tidak sesuai untuk melakukan pemodelan dan
peramalan. ARIMAX merupakan model ARIMA dengan penambahan variabel
prediktor, sedangkan Radial Basis Function Network (RBFN) merupakan model
Neural Network yang mentransformasi input secara nonlinear dengan menggunakan
fungsi aktivasi Gaussian pada lapisan unit hidden sebelum diproses secara linear pada
lapisan output. Tujuan dari penelitian ini adalah untuk mengetahui model ARIMAX,
RBFN, dan hybrid ARIMAX-RBFN serta mengetahui perbandingan akurasi dari
ketiga model tersebut. Data berasal dari Bank Indonesia. Data periode waktu 2003
sampai 2013 sebagai in-sample dan tahun 2014 sebagai out-sample. Hasil yang
diperoleh berdasarkan nilai sMAPE diketahui bahwa baik pada deret inflow maupun
outflow model ARIMAX merupakan model terbaik, dimana model tersebut adalah
ARIMAX(0,0,[1,13])(3,0,0)12 pada inflow dan ARIMAX(3,0,1)(0,0,1)12 pada
outflow. Penambahan variabel dummy pada model ARIMA mampu meningkatkan
akurasi model, terutama pada saat kejadian hari raya Idul Fitri.
=======================================================================================================
In many time series forecasting cases, the relationship between events at
different times are often not linear. Under these conditions, the approach linear
time series methods are considered not suitable for modeling and forecasting.
ARIMAX an ARIMA model with the addition of predictor variables, while the
Radial Basis Function Network (RBFN) is a model of Neural Network, which
transform inputs nonlinear using Gaussian activation function in the hidden layer
unit before being processed linearly in output layer. The purpose of this study
was to determine the ARIMA model, RBF, and hybrid ARIMAX-RBFN and
compare the accuracy of the three models. Data derived from the Indonesian
bank. Data period from 2003 through 2013 as the in-sample and 2014 as outsample. The results obtained by the value sMAPE known that both the inflow and
outflow series ARIMAX the model is the best model, where the model is
ARIMAX(0,0,[1,13])(3,0,0)12 on the inflow and ARIMAX(3,0,1)(0,0,1)12 on the
outflow. The addition of dummy variables in ARIMA model can improve the
accuracy of the model, especially at the time of the incident Eid
| Item Type: | Thesis (Masters) |
|---|---|
| Additional Information: | RTSt 519.535 Pae m |
| Uncontrolled Keywords: | ARIMAX, RBFN, Hybrid ARIMAX-RBFN, Inflow, Outflow |
| Subjects: | T Technology > T Technology (General) > T174 Technological forecasting |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis |
| Depositing User: | Mr. Tondo Indra Nyata |
| Date Deposited: | 16 Mar 2020 04:35 |
| Last Modified: | 16 Mar 2020 04:35 |
| URI: | http://repository.its.ac.id/id/eprint/75490 |
Actions (login required)
 |
View Item |