Improved Identification Of Data Correlations Through Correlation Coordinate Plots
Improved Identification Of Data Correlations Through Correlation Coordinate Plots |
Abstract
Correlation is a powerful relationship measure used in science, engineering, and business to estimate trends and make forecasts. Visualization methods, such as scatterplots and parallel coordinates, are designed to be general, supporting many visualization tasks, including identifying correlation. However, due to their generality, they do not provide the most efficient interface, in terms of speed and accuracy. This can be problematic when a task needs to be repeated frequently. To address this shortcoming, we propose a new correlation textit{task-specific} visualization method called Correlation Coordinate Plots (CCPs). CCPs transform data into a powerful coordinate system for estimating the direction and strength of correlation. To support multiple attributes, we propose 2 additional interfaces. The first is the Snowflake Visualization, a focus+context layout for exploring all pairwise correlations. The second enhances the basic CCP by using principal component analysis to project multiple attributes. We validate CCP performance in correlation-specific tasks through an extensive user study that shows improvement in both accuracy and speed.
Downloads
Citation
Hoa Nguyen, and Paul Rosen. Improved Identification Of Data Correlations Through Correlation Coordinate Plots. Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, 2016.
Bibtex
@inproceedings{nguyen2016improved, title = {Improved Identification of Data Correlations through Correlation Coordinate Plots}, author = {Nguyen, Hoa and Rosen, Paul}, booktitle = {Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications}, series = {IVAPP}, pages = {60--71}, year = {2016}, note = {textit{Best Paper Award}.}, abstract = {Correlation is a powerful relationship measure used in science, engineering, and business to estimate trends and make forecasts. Visualization methods, such as scatterplots and parallel coordinates, are designed to be general, supporting many visualization tasks, including identifying correlation. However, due to their generality, they do not provide the most efficient interface, in terms of speed and accuracy. This can be problematic when a task needs to be repeated frequently. To address this shortcoming, we propose a new correlation textit{task-specific} visualization method called Correlation Coordinate Plots (CCPs). CCPs transform data into a powerful coordinate system for estimating the direction and strength of correlation. To support multiple attributes, we propose 2 additional interfaces. The first is the Snowflake Visualization, a focus+context layout for exploring all pairwise correlations. The second enhances the basic CCP by using principal component analysis to project multiple attributes. We validate CCP performance in correlation-specific tasks through an extensive user study that shows improvement in both accuracy and speed.} }