Sklearn's Pairwise Distance Result Is Unexpectedly Asymmetrical
I am calculating the euclidean pairwise distance between elements of a vector. I use the pairwise_distances function from sklearn package. However the resulting matrix for some ele
Solution 1:
i could reproduce your error my testing for symmetry:
import numpy as np
a = np.array([[ 0.76881030949999995538490793478558771312236785888671875 ],
[ 0. ],
[ 0.67783090619999997183953155399649403989315032958984375 ],
[ 0.3228176074999999922710003374959342181682586669921875 ],
[ 0.75822395549999999087020796650904230773448944091796875 ],
[ 0.469808621599999975959605080788605846464633941650390625],
[ 0.989529862699999984698706612107343971729278564453125 ],
[ 0. ],
[ 0.5575436799999999859522858969285152852535247802734375 ],
[ 0.9756440299999999954394525047973729670047760009765625 ],
[ 0.66511863289999995085821637985645793378353118896484375 ],
[ 0.978062709200000046649847718072123825550079345703125 ],
[ 0.473957179800000016900440868994337506592273712158203125],
[ 0.82409385540000001935112550199846737086772918701171875 ],
[ 0.56548685279999999497846374651999212801456451416015625 ],
[ 0.399505730399999980928527065771049819886684417724609375],
[ 0.474232963900000026313819034839980304241180419921875 ],
[ 0.34276307189999999369689476225175894796848297119140625 ],
[ 0.9985316859999999739017084721126593649387359619140625 ],
[ 0.9063241512999999915933813099400140345096588134765625 ],
[ 0. ]])
from sklearn.metrics.pairwise import pairwise_distances
dist_sklearn = pairwise_distances(a)
print((dist_sklearn.transpose() == dist_sklearn).all())
getting False as output. Try to use scipy.spatial.distance instead. You will get a distance vector of the pairwise distance computation but can convert it to a distance matrix with squareform()
from scipy.spatial.distance import pdist, squareform
dist = pdist(a)
sq = squareform(dist)
print((sq.transpose() == sq).all())
This gave me symmetric matrix. Hope this helps
Post a Comment for "Sklearn's Pairwise Distance Result Is Unexpectedly Asymmetrical"