import pandas as pd
from sklearn.cluster import KMeans
import seaborn as sns

data_full = pd.read_csv('/datasets/cars_label.csv')

data = data_full.drop(columns=['brand'])

model = KMeans(n_clusters =3,random_state =12345)
model.fit(data)

# Дополнительный слой для центроидов
centroids = pd.DataFrame(model.cluster_centers_,columns = data.columns)
data_full['label'] = model.labels_.astype(str)
centroids['label'] = ['0 centroid', '1 centroid', '2 centroid']
data_full = pd.concat([data_full, centroids], ignore_index= True)

# Построение графика
pairgrid = sns.pairplot(data_full, hue='brand', diag_kind='hist')
pairgrid.data = centroids
pairgrid.map_offdiag(func=sns.scatterplot, s=200, marker='*', color='red')