How to do it?

Hello all,

Is there any tutorial to learn implementing a flow chart in MATLAB? By flow chart, I mean something like this:

Let´s assume I have a data set, say 24 data points , each of which 2 dimensions. Id like to cluster them in an un pre-defined number of classes. Here is the instruction:

1.FIRST: Find the center(mean) point of the whole dataset. Split this point into 2 center points. Assign each datapoint to the nearest center. 2.NOW: for each center point, repeat point 1 until none of the data points changes its membership. I think it is working in a way similar to kmeans algorithm, but the number of the classes is not defined in advance.

thank you!

[Negar]

Actually, Im stuck in the very basic part of the algorithm, I don't know how to relate each step to its previous. I have made a try, but it seems my code is not splitting the center point, it just changes its position. Here is my code:

 c = [3;2.25]; % the center point of the whole dataset
k = 1 ;
eps = 1e-3;
while k<3
    %split the centroid in 2
        c(:,k+1) = c(:,k) + eps; % the first one
        d(:,k+1) = c(:,k) - eps;  % this is not working as expected!! So how to fix it? 

    k = k+1;

end

Answer 1

[Image Analyst]

Try these links on machine learning:

http://www.mathworks.com/machine-learning/examples.html

Look at the Statistics and Machine Learning section here:

http://www.mathworks.com/discovery/?s_tid=hp_ff_a_discover

especially the section on unsupervised learning.

[Negar]

Thank you for responding, could you please have a look at my code? I can't manage how to store the center points.

[Image Analyst]

I don't know what the code does exactly but it runs without any errors at all:

c = [3;2.25]; % the center point of the whole dataset
k = 1 ;
eps = 1e-3;
while k<3
  %split the centroid in 2
  c(:,k+1) = c(:,k) + eps; % the first one
  d(:,k+1) = c(:,k) - eps;  % this works - nothing to fix.  
  k = k+1;  
end
c
d

But you should not use eps as the name of a variable since it's the name of a built in function.