How to use if, else conditional loop on anonymous function handle?

For the following example, given at https://www.mathworks.com/help/stats/evalclusters.html:

eva = evalclusters(meas, 'kmeans', 'CalinskiHarabasz', 'klist',[1:6])

Now, in the following example we use anonymous function handle:

load fisheriris
myfunc = @(X,K)(kmeans(X, K, 'emptyaction','singleton', 'replicate',5));
eva = evalclusters(meas, myfunc, 'CalinskiHarabasz', 'klist',[1:6])

kmeans returns indices, cluster center location, etc. I will like to put an if condition on myfunc before I pass it eva. So if other words, if the given condition fails, I will not like to run evalclusters.

[Adam]

Can't you just write a regular function and create a handle to that? Then you can put if statements, for loops and whatever you want in it.

[hello_world]

@ Adam: No, because eva.OptimalK is an output of evalclusters, and not of kmeans. Perhaps you mean something else. If so, please show by a working example.

[Albert Fan]

I don't quite understand you idea. The optimal number of clusters is determined by the evaluation object, and the function handle you fed in is just a clustering algorithm and that algorithm shouldn't be dependent on the optimal number of clusters. So what are you trying to achieve here?

[hello_world]

@Albert Fan: I have modified the language of my question.

Answer 1

[Albert Fan]

First of all, you cannot add a complex if statement in an anonymous function. The workaround is similar to what Adam has mentioned, is that you can create a regular function and pass the function handle in. for example:

eva = evalclusters(meas, @myfunc, 'CalinskiHarabasz', 'klist',[1:6])
function myfunc
  % code
  if condition
    %do stuff
  end
end

Additionally, the optimal cluster number is the result of evalclusters, which means you should not use that number as a condition of the clustering algorithm to stop execution. Otherwise, I personally do not know a way to stop evalcluster inside myfunc. However, if you just want to play with numbers, you can call exit() to stop the entire program. If you still want to execute something after evalclusters, then you could set a global variable and when you meet the condition, you set the variable to true and in mufunc, if that variable is true, do nothing and return. But that is rather complex and not a very good practice.