I'm trying to implementing a k -medoids clustering algorithm in Python/NumPy. As part of this algo, I have to compute the sum of distances from objects to their "medoids" (cluster representatives).
I have: a distance matrix on five points
n_samples = 5
D = np.array([[ 0. , 3.04959014, 4.74341649, 3.72424489, 6.70298441],
[ 3.04959014, 0. , 5.38516481, 4.52216762, 6.16846821],
[ 4.74341649, 5.38516481, 0. , 1.02469508, 8.23711114],
[ 3.72424489, 4.52216762, 1.02469508, 0. , 7.69025357],
[ 6.70298441, 6.16846821, 8.23711114, 7.69025357, 0. ]])
a set of initial medoids
medoids = np.array([0, 3])
and the cluster memberships
cl = np.array([0, 0, 1, 1, 0])
I can compute the required sum using
>>> np.sum(D[i, medoids[cl[i]]] for i in xrange(n_samples))
10.777269622938899
but that uses a Python loop. Am I missing some kind of vectorized idiom for computing this sum?
How about:
In [17]: D[np.arange(n_samples),medoids[cl]].sum()
Out[17]: 10.777269629999999
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.