Data relationships and their visualization

Abstract

Data analysis stands among the main directions for research in recent years. The quantity of produced data is growing at a dizzying pace, and the results of the analysis directly affect not only the operation of companies or the State but also the life of every human being. This has had a significant impact on not only existing but also future technology. Analysed data are increasingly composed of hierarchical structures with hidden links, which are hard to detect using the classic methods of data mining. This work is focused on two primary areas of data analysis. In the first a combinations of classic methods are used for the detection of hidden links in the data. We use methods for analysis of social networks, formal concept analysis, self--organising maps and others. These methods are then used with the real data, to search for experts in co--author networks, find the performance profiles of the power plant or examine the behaviour of the authors. The second area is a visualisation of obtained results, where Sammon projection is utilised to visualise the links between data with the ability to change visualisation using the different data preprocessing method.