kMeans___10___running_the_iterations.py

iterations = 100
stop=[False]*k

for n in range(iterations):
    # iterate over the data points and assign them to one of the k clusters
    print('iteration number {}'.format(n))
    
    for i, (x, y) in enumerate(zip(x_data, y_data)):      
        cluster_index= assign_cluster(x, y, kcenters)  
        # updating the color_list for all the data points on the basis of the newly assigned clusters to them.
        color_info_list[i] = color_list[cluster_index]
    
    # plotting the data points with their colors representing their newly assigned clusters
    plt.scatter(x_data, y_data, color= color_info_list)
    # plotting the k cluster centers 
    display_kcenters(kcenters)
    plt.title('k-Means iter {}'.format(n))
    plt.xlabel('Some Feature x')
    plt.ylabel('Some Feature y')
    plt.grid()
    plt.savefig(results_dir + '/iter{}.jpg'.format(n))
    plt.show()

    #calculate the new position of the centers
    for i in range(len(kcenters)):
        x_coords_list=[]
        y_coords_list= []
        for j in range(len(x_data)):
            if color_info_list[j] == color_list[i]:
                x_coords_list.append(x_data[j])
                y_coords_list.append(y_data[j])
        
        x_mean = np.mean(x_coords_list).astype(np.int32)
        y_mean = np.mean(y_coords_list).astype(np.int32)
        
        mean_shift_value = get_mean_shit_value(kcenters[i], [x_mean, y_mean])
        
        if mean_shift_value == 0:
            stop[i] = True
   
        kcenters[i] = [x_mean, y_mean]

    # check if all the cluster centers have converged
    if sum(stop) == k:
        print('all the {} cluster centers have converged on the {}th iteration'.format(k, n))
        exit()