Comparative Evaluation of Animated Scatter Plot Transitions

Scatter plots are popular for displaying 2D data, but in practice, many datasets have more than two dimensions. For the analysis of such multivariate data,it is often necessary to switch between scatter plots of different dimensionpairs, e.g., in a scatter plot matrix (SPLOM). Alternative approaches include a"grand tour" for an overview of the entire data set or creating artificial axesfrom dimensionality reduction (DR). A cross-cutting concern in all techniquesis the ability of viewers to find correspondence between data points indifferent views. Previous work proposed animations to preserve the mental mapbetween view changes and to trace points as well as clusters between scatterplots of the same underlying data set. In this paper, we evaluate a variety ofspline- and rotation-based view transitions in a crowdsourced user studyfocusing on ecological validity. Using the study results, we assess eachanimation's suitability for tracing points and clusters across view changes. Weevaluate whether the order of horizontal and vertical rotation is relevant fortask accuracy. The results show that rotations with an orthographic camera orstaged expansion of a depth axis significantly outperform all other animationtechniques for the traceability of individual points. Further, we provide aranking of the animated transition techniques for traceability of individualpoints. However, we could not find any significant differences for thetraceability of clusters. Furthermore, we identified differences by animationdirection that could guide further studies to determine potential confounds forthese differences. We publish the study data for reuse and provide theanimation framework as a D3.js plug-in.