Comparative Evaluation of Animated Scatter Plot Transitions

TL;DR

The paper tackles the challenge of tracing data correspondences across different scatter-plot views of multivariate data. It evaluates six animation techniques (three spline-based and three rotation-based) for transitions between 2D scatter plots, using a preregistered crowdsourced study with 170 participants to measure point- and cluster-tracing accuracy and subjective ratings. Rotation-based transitions, especially orthographic and staged rotation, significantly improve point-tracing performance, while cluster-tracing shows no significant differences across techniques, suggesting robust cluster correspondence across views. The authors provide a D3.js plug-in and publish the study data, offering practical guidance for integrating animated transitions into visual analytics workflows to enhance interpretability and mental map preservation.

Abstract

Scatter plots are popular for displaying 2D data, but in practice, many data sets have more than two dimensions. For the analysis of such multivariate data, it is often necessary to switch between scatter plots of different dimension pairs, 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 axes from dimensionality reduction (DR). A cross-cutting concern in all techniques is the ability of viewers to find correspondence between data points in different views. Previous work proposed animations to preserve the mental map between view changes and to trace points as well as clusters between scatter plots of the same underlying data set. In this paper, we evaluate a variety of spline- and rotation-based view transitions in a crowdsourced user study focusing on ecological validity. Using the study results, we assess each animation's suitability for tracing points and clusters across view changes. We evaluate whether the order of horizontal and vertical rotation is relevant for task accuracy. The results show that rotations with an orthographic camera or staged expansion of a depth axis significantly outperform all other animation techniques for the traceability of individual points. Further, we provide a ranking of the animated transition techniques for traceability of individual points. However, we could not find any significant differences for the traceability of clusters. Furthermore, we identified differences by animation direction that could guide further studies to determine potential confounds for these differences. We publish the study data for reuse and provide the animation framework as a D3.js plug-in.

Figures (10)

• Figure 1: Examples of spline-based animation paths for point movement in transitions between scatter plots. Gray ellipses symbolize the area covered by clusters. Time is encoded as path color (start end).
• Figure 2: 1D and 2D transitions in a SPLOM. Each square represents a cell of the SPLOM. Arrow (a) depicts a horizontal 1D transition, (b) a vertical 1D transition, and (c) a 2D transition.
• Figure 3: Staged rotation (STA) with a 3D cube happens in three sequential stages (left to right). The scatter plots at the beginning and end of the transition share one data dimension. The thick colored lines represent data dimensions during the transition.
• Figure 4: Perspective rotation (PER) has no dedicated stages (left to right). It gradually changes from an orthographic to a perspective camera and back again while simultaneously rotating the cube. The thick, colored lines represent data dimensions during the transition.
• Figure 5: User interface for the cluster task. The target cluster is only highlighted in red before the animation. After the transition, participants respond whether the cluster remained the same, merged, or split.