how to save number of organization for each cluster in k mean clustering.

i am making 4 clusters using k mean clustering. i want to get and save the indexes of the data which are a part of each cluster how can i?

%% matlab code
clc
clear all
close all
radius=15;
X0=0;    % x-coordinate of Centre of the circle
Y0=0;    % y-coordinate of Centre of circle
% Users
K=4;
t = 2*pi*rand(K,1);
R = radius*sqrt(rand(K,1));
x_u = X0 + R.*cos(t);                     % X-coordinate of the Users
y_u = Y0 + R.*sin(t);                     % Y-coordinate of the Users

X3=[-6.189; -3.251; 2.926; 6.534; 10.79; 5.203; -4.465; 1.42; -8.66; -0.8748; 6.534 ; 11.251 ;0.7298; 3; -5.23; 11.86; x_u] ;
Y3=[ 1.393; 10.42; 6.639; 5.821; 8.425; 2.924; 1.139; 0.7508; -7.917; -6.928; -10.01; -7.03; 12.48; -1.497; -6.139; -3.23; y_u ];
x_b=[0.7298; 3; -5.23; 11.86]; 
y_b=[12.48; -1.497; -6.139; -3.23]; 
P = [X3 Y3] ;   % points 
G = 4 ;   % number of groups 
C = [x_u+x_b/2 y_u+y_b/2] ;  % centroids 
idx = kmeans(P,G,'MaxIter',10,'Start',C);
gscatter(P(:,1),P(:,2),idx)

i want to store that which index value is a part of particular cluster e.g X3(2) is a part of which cluster. i want to store index 2 basically and simarly for all values

[Walter Roberson]

How would that differ from just storing the idx variable?

Caution: the "identity" of centroids can effectively exchange when using the k-means algorithm. It is not meaningful to record the centroid indices without also recording the locations of the centroids found, through the second output of kmeans()

[Maimoona Asad]

i basically have to see that which points are included in the final clusters i further need the selected ones for my tasks.

[Maimoona Asad]

[idx, C] = kmeans(P,G,'MaxIter',10,'Start',C);

are you talking about this

[Walter Roberson]

Are you asking to create cell arrays for each cluster, with just the points relevant to that cluster?

[Maimoona Asad]

i want to get the index or location of the point relevant to th cluster

[Maimoona Asad]

how can i access the location points are accessible but i want there indexes e.g the selected point in a cluster is at which index position of original data before clustering

Answer 1

[Walter Roberson]

radius=15;
X0=0;    % x-coordinate of Centre of the circle
Y0=0;    % y-coordinate of Centre of circle
% Users
K=4;
t = 2*pi*rand(K,1);
R = radius*sqrt(rand(K,1));
x_u = X0 + R.*cos(t);                     % X-coordinate of the Users
y_u = Y0 + R.*sin(t);                     % Y-coordinate of the Users
X3=[-6.189; -3.251; 2.926; 6.534; 10.79; 5.203; -4.465; 1.42; -8.66; -0.8748; 6.534 ; 11.251 ;0.7298; 3; -5.23; 11.86; x_u] ;
Y3=[ 1.393; 10.42; 6.639; 5.821; 8.425; 2.924; 1.139; 0.7508; -7.917; -6.928; -10.01; -7.03; 12.48; -1.497; -6.139; -3.23; y_u ];
x_b=[0.7298; 3; -5.23; 11.86]; 
y_b=[12.48; -1.497; -6.139; -3.23]; 
P = [X3 Y3] ;   % points 
G = 4 ;   % number of groups 
C = [x_u+x_b/2 y_u+y_b/2] ;  % centroids 

[idx, C] = kmeans(P,G,'MaxIter',10,'Start',C);

P_by_cluster = splitapply(@(varargin) varargin, P, idx);

C

C = 4×2

-1.6719 -0.3739 9.8817 -6.7567 -5.9945 -5.8297 2.8697 7.9945

P_by_cluster

P_by_cluster = 4×1 cell array
    {5×2 double}
    {3×2 double}
    {5×2 double}
    {7×2 double}