Data Science Stack Exchange is a question and answer site for Data science professionals, Machine Learning specialists, and those interested in learning more about the field. It only takes a minute to sign up.
Sign up to join this community
Is there a way to get a weight into the
pairwise_distances(X, metric='cosine')
Potentially using **kwrds?
from sklearn.metrics import pairwise_distances
In the scipy cosine distance it's possible to add in an array for weights, but that doesn't give a pairwise matrix.
a = np.array([9,8,7,5,2,9])
b = np.array([9,8,7,5,2,2])
w = np.array([1,1,1,1,1,1])
distance.cosine(a,b,w)
Where w is the weights.
Instead of using pairwise_distances you can use the pdist method to compute the distances. This will use the distance.cosine which supports weights for the values.
import numpy as np
from scipy.spatial.distance import pdist, squareform
X = np.array([[5, 4, 3], [4, 2, 1], [5, 6, 2]])
w = [1, 2, 3]
distances = pdist(X, metric='cosine', w=w)
# change the result to a square matrix
distances = squareform(distances)
Result:
array([[0. , 0.05508882, 0.04898252],
[0.05508882, 0. , 0.07833123],
[0.04898252, 0.07833123, 0. ]])
