Last active
May 17, 2021 20:49
-
-
Save Lukas0025/88d6df58fe31b64d2d36ac2935bef3b7 to your computer and use it in GitHub Desktop.
Python k-means clustering
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
## | |
# k-mean clustering algoritm | |
# @autor Lukáš Plevač <xpleva07@vutbr.cz> | |
# @date 5.5.2021 | |
# CC0 license - No Rights Reserved. | |
# | |
import numpy as np | |
import os | |
## | |
# Calcuate distance between two points (euclidean) | |
# @param point1 - np.array of point coords (same shape as of point2) | |
# @param point2 - np.array of point coords (same shape as of point1) | |
# @return distance in float | |
def euclidean_distance(point1, point2): | |
return np.sqrt(np.sum((point1 - point2) ** 2)) | |
## | |
# calculate new centroids by clusters | |
# @param clusters np.array of clusters with points | |
# @return np.array of ceteroids of clusters | |
def calc_centroids(clusters): | |
centroids = [] | |
for cluster in clusters: | |
point_len = len(cluster[0]) | |
cur_item = [] | |
for i in range(point_len): | |
cur_item.append(np.mean([point[i] for point in cluster])) | |
centroids.append(cur_item) | |
return centroids | |
## | |
# One iteration of k means clustering | |
# @param data_points - np.array of point coords to cluster | |
# @param centroids - np.array of centreal point coords (for fisrt iteration use random) | |
# @return np.array of new centroids | |
def k_means_iter(data_points, centroids, debug = False): | |
clusters = [] | |
debug_clusters = [] | |
for i in range(len(centroids)): | |
clusters.append([]) | |
debug_clusters.append([]) | |
for point_i in range(len(data_points)): | |
distance = [] | |
# clac distances to centroids | |
for centroid in centroids: | |
distance.append(euclidean_distance(data_points[point_i], centroid)) | |
# asign to cluster | |
clusters[distance.index(min(distance))].append(data_points[point_i]) | |
if debug: | |
print("point distance: " + str(distance)) | |
debug_clusters[distance.index(min(distance))].append(point_i + 1) | |
if debug: | |
print("\n\nclusters: " + str(clusters)) | |
print("\n\nclusters indexes (from 1): " + str(debug_clusters)) | |
# Calc new centroids | |
return calc_centroids(clusters) | |
## | |
# Exmaple use | |
if __name__ == "__main__": | |
data_points = np.array([[ 0,-1,-2],[-3,-1,-3],[ 1,-3, 2],[-2,-2, 2],[ 1, 2,-4],[ 0,-4, 3],[ 1, 0,-3],[-3, 0, 0],[-2, 2,-4],[-2, 4, 3],[ 3,-2, 4],[ 2,-5,-4]]) | |
centroids = np.array([[-1, 1, -4], [-1, 6, -4], [5, 0, -3]]) | |
for iteration in range(4): | |
print("\n\n---------------------------------iteration: {}".format(iteration)) | |
centroids = k_means_iter(data_points, centroids, debug=True) | |
print("\n\nnew centroids: {}".format(centroids)) | |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment