Comparison of Recently Developed Classification Tools in Microarray Data Analysis

• Activity Number: 410
• Type: Contributed
• Date/Time: Thursday, August 15, 2002 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section*
• Author(s): Jae Won Lee*+ and Jung Bok Lee and Mira Park
• Affiliation(s): Korea University and Korea University and Eulji University
• Keywords: Microarray ; discriminant analysis ; classification

Abstract

Microarray experiments allow researchers to record and identify the expression levels of thousands of genes simultaniously. A few studies cited are instances of growing body of research, in which gene expression profiling is used to distinguish between known classes of genes and to identify previously unrecognized class. In gene classification using microarray data, Dudoit et al.(2000) reviewed various classical classification methods such as LDA, Nearest Neighor, CART, and aggregating classifier methods. In addtion to classical methods, many new mehtods such as Support Vector Machine, K Nearest Neighbor using genetic algorithm and Neural Network model are developed. We investigated by simulations the performance of newly proposed methods.