ASHARI, DIMAS EWIN (2018) Penerapan Model Hybrid Singular Spectrum Analysis Deep Neural Network pada Peramalan Inflow dan Outflow Uang Kartal di Indonesia. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
06211645000006-Undergraduate_Theses.pdf - Accepted Version Download (5MB) | Preview |
Abstract
Perencanaan yang matang dalam penyusunan RKU oleh BI merupakan hal penting untuk menjaga ketersedian uang di masyarakat. Salah satu faktor yang paling diperhitungkan dalam penyusunan RKU adalah inflow dan outflow. Penelitian ini bertujuan untuk mengetahui performa ARIMAX, DNN dan SSA-DNN dalam memprediksi inflow dan outflow uang kartal kertas di Indonesia dengan data per-bulan mulai Januari 2003 - Desember 2016. Prediksi inflow dan outflow yang akurat menjadi referensi penting agar RKU dapat tersusun secara optimal sehingga kebutuhan uang dapat terpenuhi dari sisi jumlah nominal, jenis komposisi pecahan yang sesuai, tepat waktu, serta layak edar. ARIMAX, DNN dan SSA-DNN akan dibandingkan berdasarkan RMSEP dan sMAPEP dan diharapkan mampu menangkap tren, musiman dan efek variasi kalender. Pada data simulasi dengan noise linier, ARIMAX memiliki akurasi terbaik, sedangkan pada data dengan noise non-linier, DNN lebih akurat dibandingkan ARIMAX. Sementara itu, pada data inflow dan outflow DNN adalah metode terbaik pada 13 dari 14 pecahan dibandingkan ARIMAX dan SSA-DNN.
================================================================================= Mature planning in the preparation of RKU by BI is important to maintain the availability of money in the community. One of the most calculated factors in the preparation of RKU is the inflow and outflow. This study aims to determine the performance of ARIMAX, DNN and SSA-DNN in predicting inflow and outflow of currency in Indonesia with monthly data from January 2003 to December 2016. Accurate inflow and outflow prediction becomes an important reference so that RKU can be created optimally so that the need for money can be fulfilled in terms of nominal amount, the type of fractional composition, timely, and in proper condition. ARIMAX, DNN and SSA-DNN will be compared based on RMSEP and sMAPEP and are expected be able to capture the trend, seasonality and effects of calendar variations. DNN used in this research has 2 input types, DNN-1 uses significant PACF lag input from data & dummy while DNN-2 uses lag input according to AR order of ARIMAX. In the simulation data with linear noise, ARIMAX has the best accuracy, whereas in data with non-linear noise, DNN is more accurate than ARIMAX. Meanwhile, DNN application of inflow and outflow data are the best method on 13 of 14 fractions compared to ARIMAX and SSA-DNN, with 9 of them are DNN-1 type.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Uncontrolled Keywords: | ARIMAX, Deep Neural Network, Inflow, Outflow, Singular Spectrum Analysis, SSA-DNN, Variasi Kalender |
| Subjects: | Q Science > Q Science (General) > Q180.55.M38 Mathematical models Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) |
| Divisions: | Faculty of Mathematics, Computation, and Data Science > Statistics > 49201-(S1) Undergraduate Thesis |
| Depositing User: | Dimas Ewin Ashari |
| Date Deposited: | 08 Jul 2021 08:46 |
| Last Modified: | 08 Jul 2021 08:46 |
| URI: | http://repository.its.ac.id/id/eprint/57290 |
Actions (login required)
 |
View Item |