A Data Science Central Community
From chaos to clusters: origin and formation of galaxy clusters, mega cities, species, religions and everything that start as fuzzy clouds and end up in clusters. Based on mathematical modeling. Red dots represent new points, as initial points "die" and "birth" occur over time - faster at the beginning.
Done with Perl (for mathematical modeling), R (to produce 200 scatter plots) and Adobe (to assemble the scatter plots into a video). Each scatter plot has 500 dots, and required the computations of 125,000 distances, as points get attracted to neighbors. We'll publish details shortly, including source code. May converge or not depending on initial parameters and distance function. If convergence, can converge to very complex structure, a few big clusters (maybe like in this video - although the clusters change all the time), one single point (especially if edge effects are ignored), or a bunch of small clusters and super-clusters. In some cases, the evolving structure consist of small aligned clusters - it looks like a complex network of twisting filaments.
Source code: 123 lines of Perl, 10 lines of R. It takes just a few minutes to run the Perl script, the R program and playing with Adobe, to produce the video. This is indeed "small data", and I hope to do something bigger - much more points, much more scatter plots (embedded in the video), more colors, maybe 3D.
Read more at http://www.analyticbridge.com/profiles/blogs/from-chaos-to-clusters-statistical-modeling-without-models