Scientific Visualization, 2018, volume 10, number 4, pages 120 - 133, DOI: 10.26583/sv.10.4.09
Visual analysis procedures for multidimensional data
Authors: A.E. Bondarev1, A.V. Bondarenko2*, V.A. Galaktionov3
Keldysh Institute of Applied Mathematics RAS
* GOSNIIAS
1 ORCID: 0000-0003-3681-5212, bond@keldysh.ru
2 ORCID: 0000-0003-4765-6034, cod@fgosniias.ru
3 ORCID: 0000-0001-6460-7539, vlgal@gin.keldysh.ru
Abstract
The paper considers the tasks of visual analysis of multidimensional data sets. For visual analysis, an approach is used to construct elastic maps. The paper presents the results of applying elastic maps for visual analysis of multidimensional data sets of various origin. A number of data processing procedures are described, which make it possible to improve the obtained results - preliminary filtering of data, removal of separated clusters (flotation). To solve the problem of scalability, when the elastic map adjusts, both under the area of thickening data points, and to separately located points of the data cloud, the quasi-Zoom approach is applied. Examples for multidimensional data of various origins are considered - for analyzing mineral samples, for analyzing the frequencies of word sharing in word processing, for analyzing a multidimensional volume of approximation errors in the task of comparing the accuracy of solvers of a software package when modeling a flow around a cone.
Keywords: multidimensional data, visual analysis, elastic maps.
