For a Medium article

Simple k-means

from sklearn.preprocessing import StandardScaler
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import random
import seaborn as sns
from sklearn.preprocessing import StandardScaler

# We define a function returning the centroid of a set of points and the distance between two points

def centroid(df):
    return(df.mean().values)

def distance(p1,p2):
    p1=np.asarray(p1)
    p2=np.asarray(p2)
    d=np.sum((p1-p2)**2)
    return(np.sqrt(d))

# STEP 0: Scale the data
X.iloc[:,0:2] = StandardScaler().fit_transform(X)

# STEP 1: we choose a number k of clusters
k=3

# Step 2: Select k random points from the data as centroids
centroids=X.sample(k)

# Using a very simple stop criterio: 10 iterations
iter=10
for n in range(0,iter):
    # for each row we calculate the distance to each centroid
    d=[]
    for i in range(0,k):
        d.append([distance(x[1],centroids.iloc[i,:]) for x in X.iterrows()])
        
    # We assign each point to the nearest centroid
    clusters=[]
    d=np.asarray(d)
    for j in range(0,len(d[0])):
        clusters.append(np.argmin(d[0:k,j]))
    
    # We recalculate the clusters
    for m in range(0,k):
        centroids.iloc[m,:]=(centroid(X[pd.Series(clusters)==m]))