This is an implementation of this paper
import relaxed_dtw_v1
import numpy as np
dist = lambda x,y : np.abs(x-y)
x = np.array([1, 1, 2, 3, 2, 0])
y = np.array([2, 3, 3, 4, 5, 4])
distance,DTW_matrix=relaxed_dtw(x, y, distance=dist, r=3)
pip install cython
In the files, relaxed_dtw.pyx is the python code. In order to create the c extension (relaxed_dtw.c), do the following:
python setup.py build_ext --inplace
This will create the .c file and this needs to be executed everytime any change is made to the python code.
import numpy as np
import relaxed_dtw
x = np.array([1, 1, 2, 3, 2, 0])
y = np.array([2, 3, 3, 4, 5, 4])
distance,DTW_matrix=relaxed_dtw.relaxed_dtw(x,y,r =3)
- Default distance metric: (x-y)2
- Default relaxation factor "r" is 0 which is same as classic DTW