Detecting synchronization cascades in time series

I am researching delayed synchronisation in a system of coupled oscillators. There is a one-way causal connection between the oscillators, which leads to the synchronisation events occurring in a rough sequence.

When I plot this data a very clear pattern emerges:

I am looking for a way of identifying and fitting a line to the highlighted areas of the graph. Do you have any ideas on how to find these patterns?

Topic pattern-recognition

Category Data Science

1
answered at

One option is:

  1. Threshold at ~10 to remove the baseline values.
  2. Fit a clustering algorithm to identify groups.
  3. For each cluster, fit a linear regression. If the linear regression has certain a slope label cluster as a synchronization event.

About

Geeks Mental is a community that publishes articles and tutorials about Web, Android, Data Science, new techniques and Linux security.