-
Notifications
You must be signed in to change notification settings - Fork 0
/
KMeansImpl.py
47 lines (37 loc) · 1.48 KB
/
KMeansImpl.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
import numpy as np
from numpy.linalg import norm
def find_closest_cluster(distances):
return np.argmin(distances, axis=1)
class KMeansImpl:
def __init__(self, k, max_iterations=100):
self.k = k
self.max_iterations = max_iterations
self.labels = None
self.centroids = None
def initialize_centroids(self, X):
random_idx = np.random.permutation(X.shape[0])
centroids = X[random_idx[:self.k]]
return centroids
def compute_centroids(self, X, labels):
centroids = np.zeros((self.k, X.shape[1]))
for k in range(self.k):
centroids[k, :] = np.mean(X[labels == k, :], axis=0)
return centroids
def compute_distances(self, X, centroids):
distances = np.zeros((X.shape[0], self.k))
for k in range(self.k):
row_norm = norm(X - centroids[k, :], axis=1)
distances[:, k] = row_norm
return distances
def fit(self, X):
self.centroids = self.initialize_centroids(X)
for i in range(self.max_iterations):
old_centroids = self.centroids
distances = self.compute_distances(X, old_centroids)
self.labels = find_closest_cluster(distances)
self.centroids = self.compute_centroids(X, self.labels)
if np.all(old_centroids == self.centroids):
break
def predict(self, X):
distances = self.compute_distances(X, self.centroids)
return find_closest_cluster(distances)