Imagine we wanted to discover potential correlation between 2 devices, and these devices are characterized by a number of individual metrics, each having their own time series.

For example a single device might have 3 metrics like this (Device A):

enter image description here

...and another device might have its own set of readings for the same metrics (Device B):

enter image description here

How might one discover the correlation between these 2 devices? Obviously this is more complicated than looking for correlation between 2 time series since each device has multiple time series. Are there techniques for finding the correlation between 2 groups of time series?

enter image description here

  • $\begingroup$ What is the ultimate goal? Prediction? Control? ... $\endgroup$ Oct 4, 2020 at 1:11
  • $\begingroup$ @kjetilbhalvorsen To see if there is any dependency structure among devices. Are they all independent, or is their behavior correlated, such that some device depend on other devices. $\endgroup$
    – Cybernetic
    Oct 4, 2020 at 1:13
  • $\begingroup$ CANONICAL CORRELATION ANALYSIS BETWEEN TIME SERIES: arxiv.org/pdf/1210.0557.pdf $\endgroup$ Oct 4, 2020 at 1:15


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy