Visual Single Cluster of multidimensional data

Abstract

The rapid development of computer tools allows the computer system to stoke very large amount of data with many parameters such as electronic payment systems, sensors and monitoring systems and other. We talk about large data bases along both dimensions: number of recordings and number of dimensions "attribute, variable". Analysis of these data becomes very important and difficult in the same time. The visual data analysis has great potential applications because it facilitates the analysis, interpretation, validation and also increases the cognitive aspect among analysts. However, the traditional techniques of visualization of multidimensional data, such as parallel coordinates, glyphs, and scatter plot matrices, do not scale well to a very large data set. The increasing size and complexity of data sets is a new challenge and a key motivation for our works. In this article, we present our proposal approach VSCDR (Visual Single Cluster Dimension Reduction Approach) that can handle with big data.