dbscan

from math import dist
import matplotlib.pyplot as plt
from sklearn.cluster import DBSCAN
data = open('q.txt').readlines()
a = []
for e in data:
    a.append(list(map(float, e.replace(',', '.').split())))
db = DBSCAN(eps=0.27, min_samples=3)
labels = db.fit_predict(a)
plt.figure(figsize=(10, 10))
colors = ['red', 'blue', 'green', 'yellow', 'purple']
clusters = [[], [], [], [], []]
for i in range(len(a)):
    if labels[i] == -1:
        plt.scatter(a[i][0], a[i][1], color='black', s=20)
    else:
        clusters[labels[i]].append(a[i])
        plt.scatter(a[i][0], a[i][1], color=colors[labels[i]], s=20)
plt.show()
cen = []
for i in range(5):
    c_best = -1
    rst_best = 10000000000
    for c in clusters[i]:
        rst = 0
        for e in clusters[i]:
            rst += dist(e, c)
        if rst < rst_best:
            c_best = c
            rst_best = rst
    cen.append(c_best)
print(int(((cen[0][0] + cen[1][0] + cen[2][0] + cen[3][0]) / 4) * 10000), int(((cen[0][1] + cen[1][1] + cen[2][1] + cen[3][1]) / 4) * 10000))