Zayyan, Fairuz Iqbal (2026) Pemodelan Generalized Extreme Value Regression – Mixed Data Sampling Dengan Seleksi Prediktor Menggunakan Stochastic Search Variable Selection (SSVS) Untuk Prediksi Rentang Harga Impor Pakaian. Masters thesis, Institut Teknologi Sepuluh Nopember Surabaya.
![[thumbnail of 6003241027-Master_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
6003241027-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy |
Abstract
Price range merupakan mekanisme yang digunakan sebagai alat identifikasi transaksi barang impor yang disampaikan secara wajar, undervalue, atau overvalue sehingga dapat digunakan sebagai early warning system penentuan harga impor dan tarif pajak. Price range diterapkan dengan metode extreme value yang membentuk dua batas harga (atas dan bawah) sehingga dapat dimodelkan dengan menggunakan Generalized Extreme Value Regression (GEVReg). Dalam beberapa kasus ekonomi, terdapat frekuensi informasi yang berbeda antara respon dan kovariat sehingga permasalahan ini diatasi dengan proses frequency allignment pada model Mixed Data Sampling (MIDAS). Penggunaan kovariat dalam GEVReg-MIDAS membutuhkan proses seleksi untuk mengantisipasi kompleksitas model dan peningkatan beban komputasi, dimana proses seleksi diusulkan menggunakan Stochastic Search Varable Selection (SSVS) dengan mekanisme pembobotan variabel sehingga dalam penelitian diusulkan dua metode yaitu GEVReg-MIDAS dan GEVReg-MIDAS-SSVS. Penelitian ini menerapkan kedua metode pada dua data harga impor dengan kode klasifikasi 6107 dan 6109. Hasil penelitian menunjukkan bahwa GEVReg-MIDAS-SSVS lebih unggul dibandingkan GEVReg-MIDAS dengan nilai |τ-AE| dan QRMSE yang lebih kecil pada kedua batas price range. Model GEVReg-MIDAS-SSVS dipilih sebagai model terbaik dan menunjukkan kevalidan price range yang dibuktikan melalui pengujian kupiec. Pemilihan variabel menggunakan SSVS menunjukkan faktor risiko yang mempengaruhi harga impor, dimana kedua klasifikasi harga impor memiliki karakteristik faktor risiko yang berbeda antar batas price range dan mismatch-nya. Selain itu, pembobot MIDAS juga menjadi informasi faktor risiko tambahan untuk mengetahui kontribusi informasi pada bulan di kuartal. Nilai ramalan harga impor dilakukan untuk 6 periode kedepan, dimana kode 6107 berada pada interval harga Rp 597- Rp Rp 89.720,38 dan kode 6109 berada pada interval harga Rp 225,81 - Rp 1.469.712.
===================================================================================================================================
The price range is a mechanism used as a tool to identify transactions involving imported goods that are declared at a fair, undervalued, or overvalued price, so that it can be used as an early warning system for determining import prices and tax rates. Price range is applied using the extreme value method, which forms two price limits (upper and lower) so that it can be modeled using Generalized Extreme Value Regression (GEVReg). In some economic cases, there are different frequencies of information between responses and covariates, so this problem is overcome by the frequency alignment process in the Mixed Data Sampling (MIDAS) model. The use of covariates in GEVReg-MIDAS requires a selection process to anticipate model complexity and increased computational load, where the selection process is proposed using Stochastic Search Variable Selection (SSVS) with a variable weighting mechanism so that two methods are proposed in this study, namely GEVReg-MIDAS and GEVReg-MIDAS-SSVS. The results of the study show that GEVReg-MIDAS-SSVS outperforms GEVReg-MIDAS with smaller values of |τ-AE| and QRMSE at both price range limits. The GEVReg-MIDAS-SSVS model was selected as the best model and showed the validity of the price range, as proven through the Kupiec test. Variable selection using SSVS revealed the risk factors affecting import prices, where the two import price classifications had different risk factor characteristics between price range limits and mismatches. Additionally, the MIDAS weights also provided additional risk factor information to determine the contribution of information by month within a quarter. Import price forecasts were carried out for the next 6 periods, with code 6107 ranging from Rp 597 to Rp 89,720.38 and code 6109 ranging from Rp 225.81 to Rp 1,469,712.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | GEV Regression, Impor Pakaian, Mixed Data Sampling, SSVS, Clothes Import, GEV Regression, Mixed Data Sampling, SSVS |
| Subjects: | H Social Sciences > HA Statistics H Social Sciences > HA Statistics > HA30.3 Time-series analysis Q Science > QA Mathematics > QA274.2 Stochastic analysis Q Science > QA Mathematics > QA401 Mathematical models. |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis |
| Depositing User: | Fairuz Iqbal Zayyan |
| Date Deposited: | 29 Jul 2026 06:27 |
| Last Modified: | 30 Jul 2026 04:14 |
| URI: | http://repository.its.ac.id/id/eprint/139421 |
Actions (login required)
 |
View Item |