Benchmarks on time-series

[ninamiolane]

Thank you for a great package! Very clearly written, intuitive and easy to use!

Is your feature request related to a problem? Please describe. Not related to a problem.

Describe the solution you'd like Would you have benchmarks for DcorrX and MGCX on detection of correlations between time-series? Apologies in advance if there are some but I couldn't find them. Thank you!

[sampan501]

Hello, thanks for using the package! Currently we don't have benchmarks for time series correlations, but we do have a paper describing the algorithm in more detail: https://arxiv.org/pdf/1908.06486.pdf

[ninamiolane]

Thank you for your quick reply! I read this paper, very interesting too. Would you have the code for figures 1-4 maybe? Thank you so much in advance!

[sampan501]

Hi, so our code from the figures from the paper can be found here: https://github.com/neurodata/mgcpy-old/tree/j1c-time-series/demos/time_series, using the old package (now archived) and older API. I'm leaving this issue open as a reminder to port these over in the new docs

[ninamiolane]

I understand. Thank you, this is already very helpful!

[ninamiolane]

Hi again @sampan501 , I hope you are well! I was wondering if you had a linear correlation coefficient for multivariate time-series implemented? Something like RVX?

Thank you!

[sampan501]

Hi, not to my knowledge but I think @rmehta004 could help you more with that. He actually developed the theory and code for our time series module