Classification of Cancer Microarrays by v-Ridge Regression

• Activity Number: 348
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 10:30 AM to 12:20 PM
• Sponsor: International Chinese Statistical Association
• Author(s): June Luo*+
• Companies: Clemson University
• Keywords: micro-array ; cancer diagnosis ; feature selection ; support vector machine ; cross-validation

Abstract

Classification of patient samples is an important aspect of cancer diagnosis and treatment. In this paper, we propose v-ridge regression as an alternative method to the support vector machine for the unbalanced micro-array data. A primary goal in micro-array cancer diagnosis is to identify the genes which are most responsible for the classification. Two gene selection methods, univariate ranking (UR) and recursive feature elimination (RFE), are considered in the paper. Simulations on the well known Leukemia data indicate that v-ridge regression tends to select less significant genes than other methods. Simulations also demonstrate that v-ridge regression performs superior to support vector machine with lower rates in both cross-validation error and test error.