Comparing dimensionality reductions for eye movement data
Michael Burch, T.E. Kuipers, Chen Qian, Fangqin Zhou · 2020
Eye movement data is high-dimensional, and therefore hard to visualize. In this paper we focus on a dataset of scanpaths: Eye movements performed by subjects and tracked during a task which is based on path-finding. We describe comparisons of different approaches of dimensionality reduction applied to eye movement data, including t-distributed stochastic neighbor embedding (t-SNE), uniform manifold approximation and projection (UMAP), principal component analysis (PCA), and metric multidimensional scaling (MDS). We describe a tool created to analyze and compare these different methods, and perform a case study in which we explore an eye movement dataset.