Universität zu Köln

Pelchmann, Laura ORCID: 0000-0001-7933-8247 (2026). Human Perspectives on the Use of Color in Data Visualization. PhD thesis, Universität zu Köln.

PDF PhD_Thesis_Laura_Pelchmann.pdf Bereitstellung unter der CC-Lizenz: Creative Commons Attribution. Download (19MB)

Abstract

The discriminability of color differences is a critical yet under investigated factor in the effectiveness of node-link diagrams. This work consolidates five versatile investigations that examine how various visual encodings affect viewers’ ability to detect color differences in node-link diagrams and scatterplots. These visual encodings include link hues, node outlines, hue-luminance color schemes for large value ranges, semantically resonant multi-label color schemes, and factors affecting to detect qualitative differences and structural changes across graphs. Across several online experiments and controlled laboratory studies, we varied link coloration (complementary versus hue-matched), node-outline direction, outline color, and outline widths, and measured participants’ accuracy in identifying color differences between nodes. Consistently, complementary-colored links amplified node-color discriminability irrespective of underlying topology, whereas hue-similar links degraded it, while outward directed, thin black or yellow/orange outlines further enhanced discriminability for blue nodes. In a separate optimization study, we extended the Order of Magnitude Colors (OMC) scheme by systematically selecting hues and adapting luminance. Thereby we ensured color deficiency robustness, and scaling to arbitrary exponent counts. A quantitative color-metric analysis and a human-subject evaluation demonstrate that two optimized versions of the well known Glasgow scheme significantly outperform the original OMC for visualizing large value ranges (LVRs). A further investigation addresses the generation of semantically resonant color sche-mes for multi-label classes. Through two user studies and a domain validation with meteorologists, we derive three effective design strategies and implement an algorithm that produces multi-label color schemes, confirming its utility across diverse real-world datasets. Finally, a comparative study of directed acyclic graphs (DAGs) reveals that node and edge coloring, together with strategically placed uniform color regions, markedly improves the detection of structural differences across graphs under time pressure. Collectively, these findings advance the understanding of perceptual factors on color discriminability in node-link diagrams and scatterplots and inform the design of more accurate, accessible, and task-effective visualizations.

Item Type:	Thesis (PhD thesis)
Creators:	Creators Email ORCID ORCID Put Code Pelchmann, Laura laura.pelchmann@uni-koeln.de https://orcid.org/0000-0001-7933-8247 UNSPECIFIED
URN:	urn:nbn:de:hbz:38-813952
Date:	2026
Language:	English
Faculty:	Faculty of Mathematics and Natural Sciences
Divisions:	Faculty of Mathematics and Natural Sciences > Department of Mathematics and Computer Science > Institute of Computer Science
Subjects:	Data processing Computer science
Uncontrolled Keywords:	Keywords Language Visualization English Perception English Color English
Date of oral exam:	22 September 2026
Referee:	Name Academic Title von Landesberger, Tatiana Prof. Abdul-Rahman, Alfie Dr.
Refereed:	Yes
URI:	http://kups.ub.uni-koeln.de/id/eprint/81395