simple kmeans demo

kmeans_demo.py

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd

# https://drive.google.com/file/d/1ZzEouo7lRJvajxK6jLM2K_p9xAwGw1tS/view

df = pd.read_csv("clustering.csv")

k = 3
epoch = 10
X = df[["LoanAmount", "ApplicantIncome"]]
samples = X.sample(k)

def L2_dis(P1, P2):
    return ((P1.LoanAmount-P2.LoanAmount)**2 + (P1.ApplicantIncome-P2.ApplicantIncome)**2)**0.5

for _ in range(epoch):
    L2 = []
    centers = [samples.iloc[i] for i in range(k)]
    for _, p in X.iterrows():
        L2.append([L2_dis(p, c) for c in centers])

    X["c"] = np.argmin(L2, axis=1)
    samples = X.groupby("c").mean()
    print(samples)

plt.scatter(X["ApplicantIncome"], X["LoanAmount"], c='black')
plt.scatter(samples["ApplicantIncome"], samples["LoanAmount"], c='red')
plt.show()