restart a for loop when a condition is met

Hi, I want to add a loop to restart kmeans function three times when the condition is met. After running this loop, the condition is still met,then loop j will add 1 (it means that the kmeans function will be grouped j+1 clusters. )

for j = 1 : member;
  [id1,C1,sumd1,d1] = kmeans(gg{m}{i},j,'Distance','sqeuclidean');                   
  for k =1:j
      d ; %output
      if d > 1.5
         % when condition is met, restart a loop (restart kmeans function)three times.
      end
  end
end

[Jan]

Neitehr d not the condition d > 1.5 depends on the inner loop "for k". So what is the purpose of this loop? What is "member"?

[Frank Juang]

I want to add a loop to restart kmeans function three times if d >1.5.The loop j is used to run kmeans function for 1: v(member is the size of my input),so the code will group from one cluster to member clusters.

Answer 1

[KALYAN ACHARJYA]

if true
  % code
for j = 1 : member; 
  myfunction(); 
    for k =1:j; 
     if d > 1.5
      j=1; % if you want return to 1st for loop 
      %or 
      k=1; % if you want return to 2nd for loop  
      end
  end
end

[Jan]

No, you cannot modify the loop counter of a FOR loop by changing the value inside the loop. Try it:

for k = 1:10
  disp(k)
  if k == 5
     k = 8
  end
end

[Image Analyst]

Well you can but it's not good practice and only keeps the new value until the end of the loop and then goes back to what it would have been once the loop starts from the top.

If you want to change variables like loop counters inside a loop you should use a while loop instead.

Answer 2

[Jan]

What about this:

for j = 1:4
  [id1, C1, sumd1, d1] = kmeans(gg{m}{i},j,'Distance','sqeuclidean');                   
  if all(d1 <= 1.5)
    break;
  end
end

Then the loop runs either 3 times more or is stopped earlier, when the condition is met.

[Frank Juang]

Sorry, maybe I describe dully. Actually,the d in loop k is calculated by the output of kmeans function in loop j.Loop j is used to group my data from one to member clusters(member is the size of my input). My purpose is used less clusters as far as possible,so I want to restart kmeans function several times before doing j+1.If running kmeans function several times d is still large than 1.5, then j will become j+1.