Article Per Year (5 Year)
p-Index From 2020 - 2025
0.23
P-Index
This Author published in this journals
All Journal JOIV : International Journal on Informatics Visualization
Hardiyanti, Fitriani Rohmah
Unknown Affiliation
Computer Science & IT

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
Development of a Java Library with Bacterial Foraging Optimization for Feature Selection of High-Dimensional Data Badriyah, Tessy; Syarif, Iwan; Hardiyanti, Fitriani Rohmah
JOIV : International Journal on Informatics Visualization Vol 8, No 1 (2024)
Publisher : Society of Visual Informatics

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.62527/joiv.8.1.2149

Abstract

High-dimensional data allows researchers to conduct comprehensive analyses. However, such data often exhibits characteristics like small sample sizes, class imbalance, and high complexity, posing challenges for classification. One approach employed to tackle high-dimensional data is feature selection. This study uses the Bacterial Foraging Optimization (BFO) algorithm for feature selection. A dedicated BFO Java library is developed to extend the capabilities of WEKA for feature selection purposes. Experimental results confirm the successful integration of BFO. The outcomes of BFO's feature selection are then compared against those of other evolutionary algorithms, namely Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Artificial Bee Colony (ABC), and Ant Colony Optimization (ACO).  Comparison of algorithms conducted using the same datasets.  The experimental results indicate that BFO effectively reduces features while maintaining consistent accuracy. In 4 out of 9 datasets, BFO outperforms other algorithms, showcasing superior processing time performance in 6 datasets. BFO is a favorable choice for selecting features in high-dimensional datasets, providing consistent accuracy and effective processing. The optimal fraction of features in the Ovarian Cancer dataset signifies that the dataset retains a minimal number of selected attributes. Consequently, the learning process gains speed due to the reduced feature set. Remarkably, accuracy substantially increased, rising from 0.868 before feature selection to 0.886 after feature selection. The classification processing time has also been significantly shortened, completing the task in just 0.3 seconds, marking a remarkable improvement from the previous 56.8 seconds.
Co-Authors iwan Syarif Tessy Badriyah, Tessy