Eye-tracking investigation during visual analysis of projected multidimensional data with 2D scatterplots

A common strategy for visual encoding of multidimensional data for visual analyses is to use dimensionality reduction. Each multidimensional data point is projected to a 2D point using a certain strategy for the 2D layout. Many layout strategies have been proposed addressing different objectives and targeted at distinct domains and applications. The resulting projected information is typically displayed in form of 2D scatterplots. The user´s perspective such as the role of visual attention and guidance of attention for a respective layout and task has not been addressed much. It is the goal of this work to investigate, how characteristics in the layout affect the cognitive process during task completion. Eye trackers are an effective means to capture visual attention over time. We use eye tracking in a user study, where we ask users to perform typical analysis tasks for projected multidimensional data such as relation seeking, behavior comparison, and pattern identification. Those tasks often involve detecting and correlating clusters. To understand the role of point density within clusters, cluster sizes, and cluster shapes, we first conducted a study with synthetic 2D scatterplots, where we can set the respective properties manually. We evaluate how changing various parameters affect the visual attention pattern and correlate it to the correctness of the answer. In a second step, we conducted a study where the users were asked to complete tasks on real-world data with different characteristics (image collection and document collection) that are visualized using a selection of different dimensionality reduction algorithms. We transfer the insight obtained from synthetic data to investigate the decision making with real-world data. Gestalt laws can be applied to the layout structure. We examine how certain layout techniques produce certain characteristics that change the visual attention pattern. We draw some conclusions on how different projection methods support or hinder decision making leading to respective guidelines.