t-INDEPENDENT COMPONENT ANALYSIS FOR SVM CLASSIFICATION OF DNA- MICROARRAY DATA

R. AZIZ¹*, N. SRIVASTAVA², C.K. VERMA^3
1Department of Mathematics & Computer Application, Maulana Azad National Institute of Technology Bhopal - 462 051, MP, India.
²Department of Mathematics & Computer Application, Maulana Azad National Institute of Technology Bhopal - 462 051, MP, India.
³Department of Mathematics & Computer Application, Maulana Azad National Institute of Technology Bhopal - 462 051, MP, India.
* Corresponding Author : rabia.aziz2010@gmail.com

Received : 23-01-2015     Accepted : 16-03-2015     Published : 11-06-2015
Volume : 6     Issue : 1       Pages : 305 - 312
Int J Bioinformatics Res 6.1 (2015):305-312

Keywords : Independent component analysis (ICA), t-test, Support vector machine (SVM), feature selection, classification
Academic Editor : Dr S S Patil, Dr Divya Prakash Shrivastava
Conflict of Interest : None declared
Acknowledgements/Funding : The author would like to acknowledge the support of the Director (Dr. Appu Kuttan K.K.), Maulana Azad National Institute of Technology Bhopal-462051 (M.P.) India for providing basic facilities in the institute. The support of the Dr. Sanjay Sharma (Prof &

Classification analysis of microarray data is known to be hard because it involves thousands of features (genes) values, so it is necessary to reduce the number of features to obtain a manageable size of data for classification. In the present work two existing feature extraction/selection algorithms, namely Independent component analysis (ICA) and t-test are used which is a new combination of extraction/selection. The main objective of this paper is to rank the independent components of the DNA microarray data using t-test to improve the performance of Support Vector Machine (SVM)) classifier. To show the validity of the proposed method, it is applied to reduce the number of genes of five DNA microarray datasets then classify these datasets by using the SVM classifier. Experimental results on five datasets demonstrate that genes selected by proposed approach effectively improve the performance of SVM classifiers in terms of classification accuracy. We compare our proposed method with several existing methods and find that the proposed method can obtain better classification accuracy, using SVM classifier and accuracy increased up to 94.42 % of Acute leukemia data using the RBF kernel.