5

I found these two articles discussing aligning multiple 'n' number of time series using dynamic time warping or DTW.

How to apply/implement Dynamic Time Warping (DTW) or Fast Dynamic Time Warping (FastDTW) in python between 3 or more signals? Efficient pairwise DTW calculation using numpy or cython

However, they fail to describe all the steps involved to find a minimally distant path for all 'n' number of time series.

In one of the answers from the above links, it states to calculate a distance matrix of all pair-wise paths. But then, fails to mention what to do with the distance matrix in order to create a plot with 'n' number of time series that are all optimally aligned.

How would you take the distance matrix (pair-wise distances between 'n' number of time series) & align all 'n' number of time-series?

Or, is there a different approach to aligning all 'n' number of time series?

narnia649
  • 381
  • 2
  • 10
  • Did you ever find anything else on this? – Alex Luis Arias Jan 05 '22 at 19:31
  • I have tried using the tslearn library to create a barycenter average of all the trajectories. Then, I loop through all the trajectories and create a 'best match' for each to the barycenter. The pairs then are a DTW match between the original trajectory to the barycenter average. Finally, I rescaled all the trajectories to the longest trajectory match. – narnia649 Jan 20 '22 at 17:01

0 Answers0