Visualizing the distribution of Lyme diseases in 3D reported in
the U.S. in 2005, 2006, and 2007 by the CDC. One might argue that a county-level choropleth map would
be a suitable presentation. Nevertheless, it would be hard to argue that a county-level choropleth map using hue
or brightness could reveal the threefold increase in Lyme disease cases centered around New England vs. the Midwest.
Outline
Chemical Simulations
Geospatial visualizations
Multivariate data analysis
Images analysis
Matrix visualizations
Networks and structures
Scatterplot
Visual features
Scagnostics (Scatterplot Diagnostics)
Scagnostics help us to characterize 2D scatterplots
Scagnostics are computed on on three geometric graphs
A Stringy shape is a skinny shape with no branches
Computing Convex: The ratio of the area of the alpha hull and the convex hull
Example: Stringy Scagnostics
TimeSeer demo
The US Employment data comprise monthly employment rates of various economy factors for 50 states over 22 years from 1990 to 2011.
Tuan Dang and Leland Wilkinson. Transforming Scagnostics to Reveal Hidden Features. IEEE Transactions on Visualization and Computer Graphics 20(12), presented at VAST 2014
Musk dataset from UCI: https://archive.ics.uci.edu/ml/datasets/Musk+(Version+2)
Choice of transformation
The classical statistical transformations arose out of experiences applying models based on theoretical distributions to real data
ransformations we choose ought to cover the full range of negative to positive skewness as well as mixtures of distributions that are relatively symmetric
Leland Wilkinson, Anushka Anand, and Tuan Dang. CHIRP: A new classifier based on Composite Hypercubes on Iterated Random Projections. Proceedings of the 17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD 2011
Leland Wilkinson, Anushka Anand, and Tuan Dang. Substantial improvements in the set-covering projection classifier CHIRP. Journal ACM Transactions on Knowledge Discovery from Data, TKDD 2012