Visual Inference On Large P Small N Data

• Activity Number: 672
• Type: Contributed
• Date/Time: Thursday, August 4, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Graphics
• Author(s): Niladri Roy Chowdhury*+ and Dianne Cook
• Companies: Iowa State University and Iowa State University

Abstract

Statistical graphics plays an important role in exploratory data analysis, model checking and diagnosis. Recently there were some formal visual methods for determining statistical significance of findings. We often seek to low-dimensional projections in high dimensional data which reveal important aspects of the data. Projection pursuit for classification finds projections that reveal differences between classes. In this paper we are interested in the performance of classification methods when the number of observations is relatively small compared to the number of variables, known as a large p (dimension) small n (sample size) problem using visual statistical inference. We apply projection pursuit for classification to pure noise data and to the data when there is some separation. We use the lineup protocol (Buja et al. 2009) to make comparisons among the pure noise data and the data which has some separation.