cross-correlation likely not enough to capture subtle differences in sequence performance (comparing FOR to REF), investigating dynamic time warping (DTW) to enable measurement of shifting of portions of the sequence to identify distance between FOR and REF for:
overall distance metric - equivalent to temporal lag (LAG) + RMSE
warp path
bring one sequence into the space of another for direct comparison (e.g., for RMSE)
path length - likely highly correlated with overall distance metric
conceptually different since it is not the sum of the distances on the path. Test to determine if unaffected by amplitude differences.
deviation from optimal path, which should be ~LAG (but dynamic across the sequence). Again requires testing to determine if unaffected by amplitude differences
cross-correlation likely not enough to capture subtle differences in sequence performance (comparing FOR to REF), investigating dynamic time warping (DTW) to enable measurement of shifting of portions of the sequence to identify distance between FOR and REF for: