This implements the dynamic segmentation method of [REFERENCE].

This method can take any type of timestamped data, where events have some sort of distinct identities. Events should repeat, and have short-term similarity and long-term differences. This method will create segments which group time regions with similar events together. In other words, an intrinsic data scale is detected which is not related to just the event rate.

• Our paper describing the method is at [REFERENCE].
• A user manual exists at doc/Manual.rst.
• A detailed algorithmic description exists at doc/Algorithm.tex