Machine Learning by Andrew Ng --- K-means

K-means is a kind of cluster algorithm, It allows us to find characteristics in the No characteristic data.
The two important steps of K-means : One is that clustering every example to its right centroid . Another is that computing every example of each centroid’s average,then move the correspond cenroid to the average point.

Here are the whole steps of K-mean:

• random initialize K(it is better to choose K from the exit examples).

• repeat{ 1.clustering every example to its right centroid 2. computing every example of each centroid’s average,then move the correspond cenroid to the average point. }

  Something more important already mentioned in the lecture notes.

The solution (U should do it by yourself,i am just recording so i can review it future)

• findClosestCentroids.m

   
  
  minvec=zeros(K,1); 
  
  for index = 1:size(X,1) 
  
      for i = 1:K 
  
          minvec( i , 1 ) = sum( ( X(index,:)-centroids(i,:) ).^2 ) ; 
  
      end  
  
      [m , n] = min( minvec ); 
  
      idx(index , 1) = n; 
  
  end 
  
  

• computeCentroids.m

  for index = 1:K 
  
      m1 = find(idx==index); 
  
      centroids(index,:) = ( sum(X(m1,:)) )/size(m1,1); 
  
  end