Increasing the perceptual salience of relationships in parallel coordinate plots

Jonathan M. Harter, Xunlei Wu, Oluwafemi S. Alabi, Madhura Phadke, Lifford Pinto, Daniel Dougherty, Hannah Petersen, Steffen Bass, Russell M. Taylor II, Ian Roberts, Pak Chung Wong, David L. Kao (+5 others)
2012 Visualization and Data Analysis 2012  
We present three extensions to parallel coordinates that increase the perceptual salience of relationships between axes in multivariate data sets: (1) luminance modulation maintains the ability to preattentively detect patterns in the presence of overplotting, (2) adding a one-vs.-all variable display highlights relationships between one variable and all others, and (3) adding a scatter plot within the parallel-coordinates display preattentively highlights clusters and spatial layouts without
more » ... rongly interfering with the parallel-coordinates display. These techniques can be combined with one another and with existing extensions to parallel coordinates, and two of them generalize beyond cases with known-important axes. We applied these techniques to two realworld data sets (relativistic heavy-ion collision hydrodynamics and weather observations with statistical principal component analysis) as well as the popular car data set. We present relationships discovered in the data sets using these methods. Keywords Parallel coordinates; scatter plot; multivariate visualization • Luminance-Modulated Parallel Coordinates (LMPC) apply a random or variable-value-dependent luminance modulation to mitigate line cluttering and to re-introduce structure into overplotted PC plots, NIH Public Access
doi:10.1117/12.907486 pmid:23145217 pmcid:PMC3491905 dblp:conf/vda/HarterWAPPDPBT12 fatcat:feh3kgfwdvdphkcbx72bsmk5c4