ISSN 2079-3537      

 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                                                                                             

Scientific Visualization, 2018, volume 10, number 2, pages 29 - 47, DOI: 10.26583/sv.10.2.03

Visual analytics in the case of multicriteria optimization

Authors: T.P. Galkin1,A, A.P. Nelubin2,B, A.A. Galaev3,A, D.D. Popov4,A, V.V. Pilyugin5,A, S.Yu. Misyurin6,A,B

A NRNU MEPhI, Moscow, Russia

B IMASH RAN, Moscow, Russia

1 ORCID: 0000-0003-2859-6275, z@wqc.me

2 ORCID: 0000-0002-7064-3103, nelubin_andrey@inbox.ru

3 ORCID: 0000-0003-3539-3206, aalexgalaev@gmail.com

4 ORCID: 0000-0002-3333-749X, DDPopov@mephi.ru

5 ORCID: 0000-0001-8648-1690, VVPilyugin@mephi.ru

6 ORCID: 0000-0003-1020-0527, symisyurin@mephi.ru

 

Abstract

The paper considers the issue of multidimensional data analysis in the field of decision making. The authors made the formal problem statement in the general case and gave the example of the analysis for an optimization of a two-mass dynamic model. The visualization method was applied as the data analysis method. The method was shortly described. In addition, a detailed description of a visualization pipeline which was used for problem solving was given in the article. The authors formulated three types of judgments that can be made by an analyst and created the algorithm for the problem solving, which is based on the visualization method. The software tool called Visual Analytics Tool was developed. This article contains the description of this program. It is based on the visualization pipeline and allows to analyze multidimensional data using the algorithm. The results of data analysis, which was conducted with the tool, is given. They contain the description of: the sensitivity of the optimization tasks solution, the clustering process of the alternatives, which are close to an ideal, and the correlation between functional limits and objective functions values.
The study was partially supported by RFBR, research project No.16-29-04401.

 

Keywords: multidimensional data, data analysis, visualization method, multicriteria optimization.