TurbuStat Tutorials

Tutorials are provided for each of the statistic classes and their associated distance metric classes. The tutorials use the same two data sets, described on the data for tutorials page.

The plotting routines are highlighted in each of the tutorials. For users who require custom plotting routines, we recommend looking at the plotting source code as a starting point.

Distance Metrics

This section describes the distance metrics defined in Koch et al. 2017 for comparing two data sets with some output of the statistics listed above. It is important to note that few of these distance metrics are defined to be absolute. Rather, most of the metrics give relative distances and are defined only when comparing with a common fiducial image.

As shown in Koch et al. 2017, the distance metrics for some statistics have more scatter than others. Some metrics also suffer from systematic issues and should be avoided when those systematics cannot be controlled for. The Cramer distance metric is an example of this; its shortcomings are described in the paper linked above, and while the implementation is still available, we recommend caution when using it.

A distance metric for Tsallis statistics has not been explored and is not currently available in this release.