| Summary: | <p>The article deals with theoretical and practical aspects of solving the problem of visualization of multidimensional data as an effective means of multivariate analysis of systems. Several classifications are proposed for visualization techniques, according to data types, visualization objects, the method of transformation of coordinates and data. To represent classification are used charts with links to the relevant work. The article also proposes two classifications of modern trends in display technology, including integration of visualization techniques as one of the modern trends of development, along with the introduction of interactive technologies and the dynamics of development processes. It describes some approaches to the visualization problem, which are concerned with fulfilling the needs. The needs are generated by the relevant tasks such as information retrieval in global networks, development of bioinformatics, study and control of business processes, development of regions, etc. The article highlights modern visualization tools, which are capable of improving the efficiency of the multivariate analysis and searching for solutions in multi-objective optimization of technical systems, but are not very actively used for such studies. These are horizontal graphs, graphics "quantile-quantile", etc. The paper proposes to use Choropleth cards traditionally used in cartography for simultaneous presentation of the distribution parameters of several criteria in the space. It notes that visualizations of graphs in network applications can be more actively used to describe the control system. The article suggests using the heat maps to provide graphical representation of the sensitivity of the system quality criteria under variations of options (multivariate analysis of technical systems). It also mentions that it is useful to extend the supervising heat maps to the task of estimating quality of identify in constructing system models. A proposal is put forward to use the matrix scatter plots for a multivariate analysis of influence of technical system parameters on the criteria of quality, including those for reducing a dimension of the vector quality criterion and sensitivity analyses.</p><p>The paper marks not only external, but also internal relationship of visualization and optimization tasks that consists in the active use of optimization techniques in solving the problems of visualization, and vice versa. We have presented an overview of software tools, including open code, which in practice not only allow us to apply visualization techniques, but also to improve the existing baseline methodology. The conclusions are made that the end-to-end visualization systems are to be built to study specific technical objects.</p>
|
|---|
