Create a Program to Implement K Means Clustering in Python Assignment Solution.

import numpy as np

from scipy.spatial.distance import cdist

import matplotlib.pyplot as plt

data = np.loadtxt('data_points.txt', delimiter=',')

data.shape

### Define number of clusters, tolerance and maximum number of iterations

K = 4

tol = 1E-6

max_iters = 15

### Pick centroids

centroids = data[np.random.choice(data.shape[0], K),:]

print(centroids)

### Begin with K-Means Clustering Algorithm

# Define a numpy array to label each point

labels = np.zeros((data.shape[0],1))

# Define initial error

err = 1E10

iters = 0

while err > tol and iters < max_iters:

# Calculate distances to centroids

distances = cdist(data, centroids)

# Pick the minimum distance index

idx = np.argmin(distances, axis = 1)

# Now, update centroids

old_centroids = centroids.copy()

for k in range(K):

# Calculate the new value of centroid k

idxs_ = np.where(idx == k)[0]

centroid = np.mean(data[idxs_], axis = 0)

centroids[k] = centroid

labels[idxs_] = k

# Now, compute error

err = np.max(np.abs(old_centroids - centroids))

iters = iters + 1

# Print

print("Iteration {0}, error = {1:.8f}".format(iters, err))

### Plot centroids and points

plt.figure()

for k in range(K):

idxs_ = np.where(labels == k)[0]

p = data[idxs_,:]

plt.scatter(p[:,0], p[:,1], label = f"K = {k}", marker="+")

# Plot centroids

plt.scatter(centroids[:,0], centroids[:,1], color='black', marker = "x")