A new interactive tool for the classification of multivariate data through a virtual observer

A method for mapping points from a multidimensional space onto a two-dimensional plane used in conjunction with interactive graphics is presented. It provides the user with a powerful tool for visual inspection of multivariate data in the frame of unsupervised classification problems. The mapping is obtained by displaying the distance and direction of the data point, with respect to a preselected point of view and to an adjustable direction of view. A graphic user interface allows the analyst to vary the referential at his or her will, getting thus rapidly some insight into the multidimensional organisation of the data set