Authors: Jahangheer Shaik; Mohammed Yeasin; David J. Russomanno
Addresses: School of Medicine, Department of Pathology and Immunology, Washington University, St. Louis, MO 63108, USA ' Department of Electrical and Computer Engineering, University of Memphis, Memphis, TN 38152, USA ' Purdue School of Engineering and Technology, Indiana University-Purdue University, Indianapolis, IN 46202, USA
Abstract: The 3D Star Coordinate Projection (3DSCP) visualisation algorithm has been developed to address the following key issues: 1) choosing the projection configuration autonomously; 2) preserving the data topology after projection; 3) enhancing resolution. A supervised version of 3DSCP (S3DSCP) is also introduced to improve the computational efficiency of 3DSCP. Comparison with other linear, non-linear and axis-based techniques is performed to illustrate the efficacy of the 3DSCP and S3DSCP methods. Empirical analyses indicate that the 3DSCP and S3DSCP algorithms find hidden patterns in data while overcoming limitations of other techniques.
Keywords: coordinate-based projection; PCA; principal component analysis; LDA; linear discriminant analysis; LLE; locally linear embedding; exploratory data analysis; 3D star visualisation; anonymous projection configuration; data topology preservation; supervised algorithms; unsupervised algorithms; resolution enhancement.
International Journal of Data Mining and Bioinformatics, 2013 Vol.8 No.4, pp.443 - 461
Received: 15 Jan 2010
Accepted: 06 Nov 2010
Published online: 20 Oct 2014 *