tslearn.metrics.ctw

tslearn.metrics.ctw(s1, s2, max_iter=100, n_components=None, global_constraint=None, sakoe_chiba_radius=None, itakura_max_slope=None, verbose=False)[source]

Compute Canonical Time Warping (CTW) similarity measure between (possibly multidimensional) time series and return the similarity.

Canonical Time Warping is a method to align time series under rigid registration of the feature space. It should not be confused with Dynamic Time Warping (DTW), though CTW uses DTW.

It is not required that both time series share the same size, nor the same dimension (CTW will find a subspace that best aligns feature spaces). CTW was originally presented in [1].

Parameters:
s1

A time series.

s2

Another time series.

max_iter : int (default: 100)

Number of iterations for the CTW algorithm. Each iteration

n_components : int (default: None)

Number of components to be used for Canonical Correlation Analysis. If None, the lower minimum number of features between seq1 and seq2 is used.

global_constraint : {“itakura”, “sakoe_chiba”} or None (default: None)

Global constraint to restrict admissible paths for DTW calls.

sakoe_chiba_radius : int or None (default: None)

Radius to be used for Sakoe-Chiba band global constraint. If None and global_constraint is set to “sakoe_chiba”, a radius of 1 is used. If both sakoe_chiba_radius and itakura_max_slope are set, global_constraint is used to infer which constraint to use among the two. In this case, if global_constraint corresponds to no global constraint, a RuntimeWarning is raised and no global constraint is used.

itakura_max_slope : float or None (default: None)

Maximum slope for the Itakura parallelogram constraint. If None and global_constraint is set to “itakura”, a maximum slope of 2. is used. If both sakoe_chiba_radius and itakura_max_slope are set, global_constraint is used to infer which constraint to use among the two. In this case, if global_constraint corresponds to no global constraint, a RuntimeWarning is raised and no global constraint is used.

verbose : bool (default: True)

If True, scores are printed at each iteration of the algorithm.

Returns:
float

Similarity score

See also

ctw
Get only the similarity score for CTW

References

[1]F. Zhou and F. Torre, “Canonical time warping for alignment of human behavior”. NIPS 2009.

Examples

>>> ctw([1, 2, 3], [1., 2., 2., 3.])
0.0
>>> ctw([1, 2, 3], [[1., 1.], [2., 2.], [2., 2.], [3., 3.]])
0.0