Name	Natsuhiko Kumasaka
	Affiliation	Faculty of Science and Technology
	Degree	PhD in Science, Keio University
	Research Fields	Data Visualisation, Data Science
	HP	http://stat.math.keio.ac.jp/kumasaka/E/

Abstract:
The textile plot (Kumasaka and Shibata, 2006) is a cutting edge graphical display of high dimensional data which leads us to many discoveries. The textile plot is named after its appearance, where aligned warps are woven by wefts. The warps on the textile plot are self-explanatory, playing the same role as that of the axes of an observational space which can be high dimensional. Each weft which connects each coordinates on the warps signifies an observation, a point in the observational space, and scales and locations of the warps are simultaneously chosen so that all wefts are aligned as horizontally as possible. Introduction of such a horizontal alignment criterion does not only make the plot neater but also provide us a proper location of the levels of any categorical data. Two significant features of the textile plot are knots and parallel wefts (Kumasaka and Shibata, in press). Several mathematical theorems have been developed to seek for the condition for such knots or parallel wefts to be produced. The order of the warps is determined by two criteria; one is the variance criterion and another is the clustering criterion. The variance criterion indicates us several major axes useful for discriminating observations. The clustering criterion indicates us a hierarchical structure of the underlying variables.
We have implemented the textile plot as a function on the software R. This preliminary implementation suggests us the need of an integrative interactive environment for the textile plot. We have designed a reference model (Kumasaka and Shibata, 2007) for the visualisation and the interaction toward the construction of the environment. A by-product of the design is to enable us to keep track of visual interactions raised by users in a systematic way. This will lead us to further development of the environment which helps user to find better modelling.
Several practical examples are also given in Kumasaka and Shibata (2006, 2007) to show the value of the textile plot which, in turn, suggests several avenues for confirmatory data analysis despite of its high dimensionality, confirmatory data analysis despite of its high dimensionality.