To paraphrase the adage, a picture is worth a thousand numbers. In order to investigate some basic properties of NSRR datasets, here we generate a number of whole-dataset visualizations. To make sense of these images, we’ll employ a remarkably complex computational pattern recognition and dimension reduction framework, a.k.a. the human visual system. We'll illustrate how visualization can highlight temporal, ultradian patterns in sleep across many individuals; more generally, we'll also see how it can point to oddity and artifact in datasets, as well as acting as an hypothesis-generating tool, especially for complex, multi-modal datasets.
The full post can be viewed as a vignette on the Luna website:
URL : [http://zzz.bwh.harvard.edu/luna/vignettes/dataplots/]