Dear list,
I have a collection of hundreds of 2d scatterplots. In fact, I don't have actual graphics, but I have hundreds of (x,y) planes. I would like to use graph-theoretic diagnostics to detect abnormalities or patterns within each x,y plane. The method I would like to use was introduced by John and Paul Tukey in the 1990s, who achieve my objective through a small number of measures. Dang and Wilkinson (see http://www.cs.uic.edu/~tdang/file/Ti...NCIjournal.pdf) describe in an excellent fashion on how one can use these so-called "Scagnostics" to explore abnormalities in huge amounts of (x,y) planes. Finally, Wilkinson, Anand, and Grossman [1] provide a detailed description on the idea of graph-theoretic scagnostics. These scagnostics or graph-theoretic measures are exactly what I need.
More specifically, Wilkinson et al. suggest to compute the following measures for each of my (x,y) planes: Outlying, Skewed, Clumpy, Sparse, Striated, Convex, Skinny, Stringy, Monotonic. They describe each of these "scagnostics" in their paper. I would like to compute each and every one of these measures, for hundreds of (x,y) planes. I do not need my result in graphs, rather I need the actual numeric measures (scagnostics).
However, before I try to come up with efficient ways on computing each metric, does anyone know of a user-written program that does what I'm looking for?
Thank you very much for your help.
[1] Wilkinson, Leland, Anushka Anand, and Robert L. Grossman. "Graph-Theoretic Scagnostics." INFOVIS. Vol. 5. 2005.
I have a collection of hundreds of 2d scatterplots. In fact, I don't have actual graphics, but I have hundreds of (x,y) planes. I would like to use graph-theoretic diagnostics to detect abnormalities or patterns within each x,y plane. The method I would like to use was introduced by John and Paul Tukey in the 1990s, who achieve my objective through a small number of measures. Dang and Wilkinson (see http://www.cs.uic.edu/~tdang/file/Ti...NCIjournal.pdf) describe in an excellent fashion on how one can use these so-called "Scagnostics" to explore abnormalities in huge amounts of (x,y) planes. Finally, Wilkinson, Anand, and Grossman [1] provide a detailed description on the idea of graph-theoretic scagnostics. These scagnostics or graph-theoretic measures are exactly what I need.
More specifically, Wilkinson et al. suggest to compute the following measures for each of my (x,y) planes: Outlying, Skewed, Clumpy, Sparse, Striated, Convex, Skinny, Stringy, Monotonic. They describe each of these "scagnostics" in their paper. I would like to compute each and every one of these measures, for hundreds of (x,y) planes. I do not need my result in graphs, rather I need the actual numeric measures (scagnostics).
However, before I try to come up with efficient ways on computing each metric, does anyone know of a user-written program that does what I'm looking for?
Thank you very much for your help.
[1] Wilkinson, Leland, Anushka Anand, and Robert L. Grossman. "Graph-Theoretic Scagnostics." INFOVIS. Vol. 5. 2005.
Comment