import math
import numpy as np
import matplotlib.pyplot as plt
from random import randint

def distance(cluster, point):
    return math.sqrt( ((cluster[0] - point[0]) ** 2) + ((cluster[1] - point[1]) ** 2) )

def distances_from_center(center, data):
    distances = []
    for p in dataset:
        # voor elk punt de afstand tot het huidige cluster centrum
        distances.append(distance(c, p))
    return distances

def calculate_new_center(cluster):
    new_center_x = sum([row[0] for row in cluster]) / len(cluster)
    new_center_y = sum([row[1] for row in cluster]) / len(cluster)
    return [new_center_x, new_center_y]

dataset = []
for i in range(0,100000):
    dataset.append([randint(0,500), randint(0,500)])
cluster_count = 5
cluster_centers = []
clusters = []

for i in range(cluster_count):
    cluster_centers.append(dataset[i])
    clusters.append([])

done = False

last_distance_matrix = []
while not done:
    distance_matrix = []
    for i in range(0, len(clusters)):
        clusters[i] = []

    # Voor elke cluster centrum de afstanden berekenen
    # Dit kan in parallel
    for c in cluster_centers:
        distance_matrix.append(distances_from_center(c, dataset))

    # Distance_matrix omzetten naar een numpy array
    distance_matrix = np.array(distance_matrix)

    # Datapunten toewijzen aan cluster_centers
    # Dit kan in parallel
    for p in range(0, len(distance_matrix[0])):
        min_index = np.argmin(distance_matrix[:, p])
        clusters[min_index].append(dataset[p])

    # Nieuwe cluster_centers berekenen
    # Kan in parallel
    for x in range(0, len(cluster_centers)):
        cluster_centers[x] =  calculate_new_center(clusters[x])

    if np.all(distance_matrix == last_distance_matrix):
        print("done")
        done = True
    else:
        print("not done")
        last_distance_matrix = distance_matrix
for c in clusters:
    x_data = [r[0] for r in c]
    y_data = [r[1] for r in c]
    plt.scatter(x_data, y_data)

x_clusters = [r[0] for r in cluster_centers]
y_clusters = [r[1] for r in cluster_centers]
plt.scatter(x_clusters, y_clusters)
plt.show()