Question

I need this code to be written in Python:

Given a dataset, D, which consists of (x,y) pairs, and a list of cluster assignments, C, write a function centroids(D, C) that computes the new (x,y) centroid for each cluster. Your function should return a list of the new cluster centroids, with each centroid represented as a list of x, y:

def centroid(D, C):

Answer 1

The function to get the new centroid was not given, so I took it as average of all data points associated with that cluster. (k-means)

Program:

def centroid(D, C):
  
   # the list of new centroids
   new_centroids = []
   # number of data points in a cluster (associated with a centroid)
   number_of_elements_in_cluster = []
   # number of clusters -- starting from 1 to n
   max_cluster_number = 0
  
   # number of data points
   n1 = len(D)
  
   # get number of clusters by getting the max cluster number
   for i in range(0,n1):
       if(C[i]>max_cluster_number):
           max_cluster_number = C[i]
  
   # initialze the new centroids as 0,0
   # and the number of elements in each cluster = 0
   for i in range(0, max_cluster_number):
       new_centroids.append([0,0])
       number_of_elements_in_cluster.append(0)
      
   # get the sum of data points in each cluster
   # add that data point in the number of elements for that cluster
   for i in range(0,n1):
       new_centroids[C[i]-1][0] += D[i][0]
       new_centroids[C[i]-1][1] += D[i][1]
       number_of_elements_in_cluster[C[i]-1] += 1
      
   # Divide each cluster sum by number of data points to get the means
   # which is the new centroid
   for i in range(0, max_cluster_number):
       new_centroids[C[i]-1][0] /= number_of_elements_in_cluster[C[i]-1]
       new_centroids[C[i]-1][1] /= number_of_elements_in_cluster[C[i]-1]
  
   return new_centroids

# sample run
D = [[2,0],[3,5],[4,6]]
C = [2,1,1]
print(centroid(D,C))