Exploring feature selection method for microarray classification

Abstrak

Effectively selecting features from high-dimensional microarray data is essential for accurate cancer detection. This study explores the pivotal role of feature selection in improving the accuracy of classifying microarray data for ovarian cancer detection. Utilizing machine learning techniques and microarray technology, the research aims to identify subtle gene expression patterns that indicate ovarian cancer. The research explores the utilization of principal component analysis (PCA) for dimensionality reduction and compares the effectiveness of feature selection techniques such as artificial bee colony (ABC) and sequential forward floating selection (SFFS). The dataset used in this study comprises of 15154 genes, 253 instances, and 2 classes related to ovarian cancer. Through a comprehensive analysis, the study aims to optimize the classification process and improve the early detection of ovarian cancer. Moreover, the study presents the classification accuracy results obtained by PCA, ABC, and SFFS. While PCA achieved an accuracy of 96% and SFFS yielded a classification accuracy of 98%, ABC demonstrated the highest classification accuracy of 100%. These findings underscore the effectiveness of ABC as the preferred choice for feature selection in improving the classification accuracy of ovarian cancer detection using microarray data.