amunozj / fastdtw

A Python implementation of FastDTW

fastdtw

Python implementation of FastDTW¹, which is an approximate Dynamic Time Warping (DTW) algorithm that provides optimal or near-optimal alignments with an O(N) time and memory complexity.

Install

pip install fastdtw

Example

import numpy as np
from scipy.spatial.distance import euclidean

from fastdtw import fastdtw

x = np.array([[1,1], [2,2], [3,3], [4,4], [5,5]])
y = np.array([[2,2], [3,3], [4,4]])
distance, path = fastdtw(x, y, dist=euclidean)
print(distance)

Example parallel computing -------

import numpy as np
from scipy.spatial.distance import euclidean

from fastdtw import fastdtw_parallel, get_path

X = np.random.randint(1, 40, size=(5, 5))
distance_matrix, path_list = fastdtw_parallel(X, dist=euclidean, n_jobs=-1)
print(distance_matrix)

# To allow the straight forward access of warp paths to the respective distance matrix element
# the get_path function can be used
# Access the path for distance matrix element with index row=2 and column=3
print(get_path(path_list, 2, 3)

References

---

1. Stan Salvador, and Philip Chan. "FastDTW: Toward accurate dynamic time warping in linear time and space." Intelligent Data Analysis 11.5 (2007): 561-580.↩