least square density difference for "change point" detection

[Datseris]

Indicator summary

Alright, I just saw this Julia code, https://github.com/johncwok/ChangePointDetection.jl , that implements an algorithm called least square density difference (LSDD) method. It is used to detect changepoints in time-series or to infer wether or not two time-series come from the same underlying probability distribution.

Reference

The LSDD method was developped in the article Density-difference estimation from M. Sugiyama, T. Suzuki, T. Kanamori, M. C. du Plessis, S. Liu, and I. Takeuchi. in 2013.

http://www.ms.k.u-tokyo.ac.jp/sugi/2013/LSDD.pdf

Codebase

https://github.com/johncwok/ChangePointDetection.jl

Implementation plan

Best to have a look at the code and see if it can be improved performance wise. Probably also integrated with our sliding window viewer.