z-logo
open-access-imgOpen Access
ANALISIS METODE RBF-NN DENGAN OPTIMASI ALGORITMA GENETIKA PADA PERAMALAN MATA UANG EUR/USD
Author(s) -
Nengah Widiangga Gautama,
Agus Dharma,
Made Sudarma
Publication year - 2016
Publication title -
majalah ilmiah teknologi elektro
Language(s) - English
Resource type - Journals
eISSN - 2503-2372
pISSN - 1693-2951
DOI - 10.24843/mite.1502.16
Subject(s) - computer science
This paper discuss about EUR/USD forex forecasting using RBF-NN (Radial Basis Function – Neural Network) method without optimization and RBF-NN optimized by 3models of Genetic Algorithm (GA) and Adaptive Genetic Algorithm (AGA). Forecasting, which is done with the help of MATLAB program, uses 2 types of data (daily high and daily low) and 2 to 5 clusters model. The accuracy is shown as MAPE (Mean Absolute Percentage Error) value. In daily low data, method with the best MAPE is GA type II (3 clusters) with MAPE value of 0.2286%, while in daily high data, method with the best MAPE is AGA type II (3 clusters) with MAPE value of 0.2190%. GA type II and AGA type II has certain technique which searches around RBF-NN's weight, and this technique proofed to be effective in EUR/USD case. Improvement in accuracy that GA type II and AGA type II gives to RBF-NN method may also be used in other currency as well. Intisari—Penelitian ini membahas tentang peramalan EUR/USD menggunakan metode RBF-NN (Radial Basis Function – Neural Network) tanpa optimasi dan RBF-NN yang dioptimasi dengan 3 model AG/AGA (Algoritma Genetika dan Algoritma Genetika Adaptif). Sistem RBF-NN dapat diterapkan pada data dengan karakteristik nonlinear dan fluktuatif seperti data EUR/USD. Permasalahan akurasi muncul jika terjadi solusi lokal dalam sistem RBF-NN dan metode AG/AGA dapat digunakan untuk mengatasi solusi lokal tersebut. Keakuratan dari peramalan ditunjukkan lewat nilai MAPE (Mean Absolute Percentage Error). Pada data daily low, metode terbaik adalah Algoritma Genetika II dengan MAPE sebesar 0,2286%, sementara pada data daily high metode terbaik adalah Algoritma Genetika Adaptif II dengan MAPE sebesar 0,2190%. Metode AG II dan AGA II didukung teknik pencarian di dekat bobot RBFNN yang terbukti efektif pada kasus mata uang EUR/USD. Perbaikan akurasi yang diberikan AG II dan AGA II terhadap metode RBF-NN dapat diterapkan pada peramalan mata uang lainnya. Kata Kunci— EUR/USD, RBF-NN, algoritma genetika, algoritma genetika adaptif, cluster, MAPE, daily high, daily low.

The content you want is available to Zendy users.

Already have an account? Click here to sign in.
Having issues? You can contact us here
Accelerating Research

Address

John Eccles House
Robert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom