This paper presents a guideline for visualization designers who want to choose appropriate techniques for enhancing tasks involving multidimensional projection. Specifically, we adopt a user-centric approach in which we take user perception into consideration. Here, we focus on projection techniques that output 2D or 3D scatterplots that can then be used for a range of common data analysis tasks, which we categorize as pattern identification tasks, relation-seeking tasks, membership disambiguation tasks, or behavior comparison tasks. Our user-centric task categorization can be used to effectively guide the organization of multidimensional data projection layouts. Moreover, we present real-world examples that demonstrate effective choices made by visualization designers faced with complex datasets requiring dimensionality reduction.

Keywords: Multidimensional data analysis, task taxonomy, multidimensional data projection, user-centric evaluation.

PDF

Etemadpour, R., Linsen, L., Paiva, J. G., Crick, C., Forbes, A. G., Choosing Visualization Techniques for Multidimensional Data Projection Tasks: A Guideline with Examples, Computer Vision, Imaging and Computer Graphics – Theory and Applications, Communications in Computer and Information Science, Springer, January 1st, 2016.