Publications

Sunspot Plots: Model-based Structure Enhancement for Dense Scatter Plots

T. Trautner, F. Bolte, S. Stoppel, and S. Bruckner

Abstract

Scatter plots are a powerful and well-established technique for visualizing the relationships between two variables as a collection of discrete points. However, especially when dealing with large and dense data, scatter plots often exhibit problems such as overplotting, making the data interpretation arduous. Density plots are able to overcome these limitations in highly populated regions, but fail to provide accurate information of individual data points. This is particularly problematic in sparse regions where the density estimate may not provide a good representation of the underlying data. In this paper, we present sunspot plots, a visualization technique that communicates dense data as a continuous data distribution, while preserving the discrete nature of data samples in sparsely populated areas. We furthermore demonstrate the advantages of our approach on typical failure cases of scatter plots within synthetic and real-world data sets and validate its effectiveness in a user study.

T. Trautner, F. Bolte, S. Stoppel, and S. Bruckner, "Sunspot Plots: Model-based Structure Enhancement for Dense Scatter Plots," Computer Graphics Forum, vol. 39, iss. 3, p. 551–563, 2020. doi:10.1111/cgf.14001
[BibTeX]

Scatter plots are a powerful and well-established technique for visualizing the relationships between two variables as a collection of discrete points. However, especially when dealing with large and dense data, scatter plots often exhibit problems such as overplotting, making the data interpretation arduous. Density plots are able to overcome these limitations in highly populated regions, but fail to provide accurate information of individual data points. This is particularly problematic in sparse regions where the density estimate may not provide a good representation of the underlying data. In this paper, we present sunspot plots, a visualization technique that communicates dense data as a continuous data distribution, while preserving the discrete nature of data samples in sparsely populated areas. We furthermore demonstrate the advantages of our approach on typical failure cases of scatter plots within synthetic and real-world data sets and validate its effectiveness in a user study.
@article{Trautner-2020-SunspotPlots,
author = {Trautner, T. and Bolte, F. and Stoppel, S. and Bruckner, S.},
title = {Sunspot Plots: Model-based Structure Enhancement for Dense Scatter Plots},
journal = {Computer Graphics Forum},
volume = {39},
number = {3},
pages = {551--563},
keywords = {information visualization, scatterplots, kernel density estimation},
doi = {10.1111/cgf.14001},
abstract = {Scatter plots are a powerful and well-established technique for visualizing the relationships between two variables as a collection of discrete points. However, especially when dealing with large and dense data, scatter plots often exhibit problems such as overplotting, making the data interpretation arduous. Density plots are able to overcome these limitations in highly populated regions, but fail to provide accurate information of individual data points. This is particularly problematic in sparse regions where the density estimate may not provide a good representation of the underlying data. In this paper, we present sunspot plots, a visualization technique that communicates dense data as a continuous data distribution, while preserving the discrete nature of data samples in sparsely populated areas. We furthermore demonstrate the advantages of our approach on typical failure cases of scatter plots within synthetic and real-world data sets and validate its effectiveness in a user study.},
year = {2020},
pdf = "pdfs/Trautner_2020_SunspotPlots_PDF.pdf",
thumbnails = "images/Trautner_2020_SunspotPlots_thumb.png",
images = "images/Trautner_2020_SunspotPlots_thumb.png",
vid = "vids/Trautner_2020_SunspotPlots_video.mp4",
youtube = "https://youtu.be/G6l-y6YGjzQ",
project = "MetaVis"
}
projectidMetaVisprojectid

Media

Downloads

Full paper [PDF]