Metode Hibridasi Artificial Bee Colony dan Fuzzy K-Modes untuk Klasterisasi Data Kategorikal
| Main Author: | Khalid, Khalid |
|---|---|
| Format: | Article info application/pdf eJournal |
| Bahasa: | eng |
| Terbitan: |
Program Studi Sistem Informasi Fakultas Sains dan Teknologi, UIN Sunan Ampel Surabaya
, 2019
|
| Subjects: | |
| Online Access: |
http://jurnalsaintek.uinsby.ac.id/index.php/SYSTEMIC/article/view/466 http://jurnalsaintek.uinsby.ac.id/index.php/SYSTEMIC/article/view/466/346 |
Daftar Isi:
- Fuzzy K-Modes is an effective method for clustering categorical data. This method is as extensions of fuzzy k-means algorithm by using modes in the process of matching the dissimilarity measure to update centroid of the cluster and to obtain the optimal solution. Nevertheless, Fuzzy K-Modes has the disadvantage of the possibility of stopping in the optimal local solution. Artificial Bee Colony (ABC) is an optimization method that has been proven effective and has the ability to obtain global solutions. This study proposes a hybridization between the Artificial Bee Colony algorithm and Fuzzy K-Modes for clustering categorical data. The implementation of hybridization between Artifical Bee Colony and Fuzzy K-Modes (ABC-FKMO) has been proven to be able to improve the performance of categorical data clustering especially in the aspects of Objective Function, F-Measure, and Accuracy. The test results with datasets of the Soybean Disease, Breast Cancer and Congressional Voting Records from the UCI data repository, showed the Accuracy averages of 0.991, 0.615, and 0.867. Objective Function is better at an average of 2.73%, F-Measure is better at an average of 4.31% and Accuracy is better at an average of 5.16%.
- Fuzzy K-Modes merupakan metode klasterisasi data yang efektif untuk data kategorikal. Metode ini merupakan perluasan fuzzy k-means dengan menggunakan modes (modus) dalam proses pencocokan ukuran ketidaksamaan (dissimilarity measure) untuk memutakhirkan titik pusat klaster dan mendapatkan solusi yang optimal. Meskipun demikian Fuzzy K-Modes memiliki kelemahan adanya kemungkinan berhenti dalam solusi lokal optimal. Artificial Bee Colony (ABC) merupakan metode optimasi yang sudah terbukti efektif dan memiliki kemampuan mendapatkan solusi global. Penelitian ini mengusulkan hibridasi algoritma Artificial Bee Colony dengan Fuzzy K-Modes untuk klasterisasi data kategorikal. Implementasi hibridasi Artifical Bee Colony dengan Fuzzy K-Modes (ABC-FKMO) terbukti mampu meningkatkan performa klasterisasi data kategorikal khususnya dalam aspek nilai Objective Function, F-Measure, dan Accuracy. Hasil pengujian dengan dataset Soybean Disease, Breast Cancer dan Congressional Voting Records dari UCI data repository, menunjukkan rata-rata Accuracy sebesar 0.991, 0.615, dan 0.867. Objective Function lebih baik rata rata sebesar 2,73 %, F-Measure lebih baik rata-rata sebesar 4,31 % dan Accuracy lebih baik rata-rata sebesar 5,16 %.