Articles

VOTING OF ARTIFICIAL NEURAL NETWORK PARTICLE SWARM OPTIMIZATION BICLASSIFIER USING GAIN RATIO FEATURE SELECTION

Main Article Content

Fetty Tri Anggraeny Monica Widiasri

Abstract

VOTING OF ARTIFICIAL NEURAL NETWORK PARTICLE SWARM OPTIMIZATION BICLASSIFIER USING GAIN RATIO FEATURE SELECTION a Fetty Tri Anggraeny, bMonica Widiasri aTeknik Informatika Universitas Pembangunan Nasional “Veteran” Jawa Timur Jl. Raya Rungkut Madya Gunung Anyar Surabaya Indonesia aUniversitas Surabaya Jawa Timur Jl. Raya Kalirungkut Surabaya Indonesia E-Mail: fetty_ta@yahoo.com Abstrak Seleksi fitur merupakan tahapan penting dalam proses klasifikasi. Proses ini menganalisa data (fitur) sehingga menghasilkan fitur yang berperan atau kurang berperan dalam proses klasifikasi. Fitur yang kurang berperan dapat tidak digunakan dalam proses klasifikasi. Peranan sebuah fitur dalam klasifikasi dapat dikalkulasi dengan suatu rumusan, dalam penelitian ini digunakan metode gain ratio untuk mendapatkan bobot atribut dalam proses klasifikasi. Gain ratio pengembangan dari information gain yang digunakan untuk membangun pohon keputusan (decision tree). Metode seleksi fitur gain ratio menggunakan pendekatan seleksi fitur filter, karena dilakukan terlepas dari mesin klasifikasi. Mesin klasifikasi yang digunakan adalah Artificial Neural Network Particle Swarm Optimization (ANNPSO), dimana mesin ini menggabungkan konsep kecerdasan buatan saraf manusia (neural networks) dengan kecerdasan hewan (particle swarm intelligence). Metode yang diusulkan akan diuji coba terhadap 3 dataset UCI, antara lain iris, breast Wisconsin dan dermatology. Uji coba dengan variasi nilai batas gain ratio fitur menunjukkan nilai akurasi yang cukup tinggi terhadap 3 dataset yaitu 97,6%, 96,41%, dan 99,29%. Kata kunci: Gain Ratio, klasifikasi suara terbanyak, ANNPSO biclassifier.. Abstract Feature selection is an important step in classification process, it analyze the data (features) resulting role each features in the classification process. The role of a feature in the classification can be calculated with a formula, in this research the gain ratio method is used to get the attribute/feature weights. Gain ratio is the development of information gain. Information gain is used to form the induction of decision tree (ID3). Gain ratio feature selection method using the filter feature selection approach, as is done separately from classification engine. Classification engine used is Voting of Artificial Neural Network Particle Swarm Intelligence (ANNPSO) Biclassifier, where this engine combines the concept of artificial intelligence human nerve (neural network) with animal intelligence (particle swarm intelligence). The proposed method is tested on three datasets of UCI, including iris, breast wisconsin and dermatology. Trials with the variation of the boundary gain ratio feature showed a high accuracy of the three datasets are 97.6%, 96.41%, and 99.29%. Keywords: Gain Ratio, Voting Classification, ANNPSO Biclassifier

Downloads

Download data is not yet available.

Article Details

How to Cite
ANGGRAENY, Fetty Tri; WIDIASRI, Monica. VOTING OF ARTIFICIAL NEURAL NETWORK PARTICLE SWARM OPTIMIZATION BICLASSIFIER USING GAIN RATIO FEATURE SELECTION. Jurnal Ilmiah Kursor, [S.l.], v. 7, n. 2, july 2013. ISSN 2301-6914. Available at: <https://kursorjournal.org/index.php/kursor/article/view/44>. Date accessed: 17 dec. 2018.
Section
Articles