| Abstrakt | | Informative data visualisation is an essential part of modern statistical data analysis, and thus of research and professional practice in biomedicine and elsewhere. The thesis makes an original contribution in two fields of data visualisation. The first part is aimed at improving visualisation of large amounts of three-dimensional data consisting of one categorical and two numeric dimensions by means of augmented convex hull plots. The second part is dedicated to the previously unaddressed task of visualising concordance, with emphasis on comparison of concordance between groups of raters. The newly proposed methods were implemented in open-source public-domain or widely available software working on various platforms, and tested on real-life datasets from biostatistical practice. Feasibility of augmented convex hull plots and the software to produce them was additionally assessed with a survey of potential users. The proposed diagrams were described on the basis of Wilkinson's grammar of graphics. The proposed data visualisation methods can contribute to better insight into studied phenomena, and thus improve quality and clarity of publications.
|