Kurniawan, Ade (2017) Prediksi Tren Kurs Dollar Dari Berita Finansial Amerika Serikat Berbahasa Indonesia Menggunakan Support Vector Machine. Sarjana thesis, Universitas Brawijaya.
Abstract
Dolar Amerika Serikat (USD) merupakan mata uang yang paling sering digunakan dalam transaksi internasional dengan sirkulasi harian terbesar dibandingkan mata uang negara lain. Rilis data finansial Amerika Serikat tidak hanya berdampak terhadap perekonomian negara itu sendiri tetapi juga akan memberikan dampak terhadap perekonomian negara-negara lain. Fokus utama dari penelitian ini adalah prediksi tren kurs USD dengan cara klasifikasi berita finansial Amerika Serikat berbahasa Indonesia menggunakan algoritme Support Vector Machine (SVM). Kernel yang digunakan pada penelitian ini adalah polynomial degree d, komposisi data terbaik ketika menggunakan rasio 80% untuk data training dan 20% untuk data testing. Output yang dihasilkan dibagi menjadi 2 kelas yakni berdampak melemahkan (down) atau yang berdampak menguatkan (up) kurs USD terhadap mata uang rival. Adapun pengujian terbaik diperoleh dengan kombinasi parameter DF threshold bawah = 15%, DF threshold atas = 85%, λ=0.1, CLR=0.01, C=1, epsilon=0.00001, dan iterasi maksimal =100. Dari hasil pengujian dihasilkan akurasi rata-rata sebesar 76.66%, nilai sensitivitas 80% dan spesifisitas 73.33%.
English Abstract
United States Dollar (USD) is the most used currency for international transaction, its daily circulation is bigger than the other currency in the world. America's financial data not only give an impact in America itself but also directly effecting the other country. The main focus of this research is to predict USD's trend from America's Financial News in Bahasa Indonesia Using Support Vector Machine Algorithm. Kernel that used in this research is polynomial degree d, the best data composition when we using 80% for training data and 20% for testing data. The output generated into 2 class to weaken USD price (Down) and on the other hand to strengthen USD price (Up) to rival's currency. The best parameter combination that give best average accuracy are using under DF threshold = 15%, upper DF threshold = 85%, λ=0.1, CLR=0.01, C=1, epsilon=0.00001, maximum iteration=100 and generated average accuracy=76.66%, sensitivity=80% and specificity=73.33%.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | SKR/FTIK/2017/423/051707745 |
| Uncontrolled Keywords: | Forex, Klasifikasi, Support Vector Machine |
| Subjects: | 000 Computer science, information and general works > 006 Special computer methods > 006.3 Artificial intelligence > 006.31 Machine learning |
| Divisions: | Fakultas Ilmu Komputer > Teknik Informatika |
| Depositing User: | Yusuf Dwi N. |
| Date Deposited: | 22 Sep 2017 08:54 |
| Last Modified: | 05 Nov 2020 03:39 |
| URI: | http://repository.ub.ac.id/id/eprint/2866 |
![[thumbnail of BAB I.pdf]](http://repository.ub.ac.id/style/images/fileicons/text.png) |
Text
BAB I.pdf Restricted to Repository staff only Download (310kB) |
|
Text
BAB II.pdf Restricted to Repository staff only Download (477kB) |
|
Text
BAB III.pdf Restricted to Repository staff only Download (530kB) |
|
Text
BAB IV.pdf Restricted to Repository staff only Download (387kB) |
|
Text
BAB V.pdf Restricted to Repository staff only Download (642kB) |
|
Text
BAB VI.pdf Restricted to Repository staff only Download (374kB) |
|
Text
BAB VII.pdf Restricted to Repository staff only Download (282kB) |
|
Text
Bagian Depan.pdf Restricted to Repository staff only Download (362kB) |
|
Text
Daftar Pustaka.pdf Restricted to Repository staff only Download (281kB) |
Actions (login required)
 |
View Item |