Reducing repeated elements in an array by a factor

Is there a way to reduce the number of repeated elements in an array by a factor? For example, say the factor I had was 3, i would want to reduce

[1 2 2 2 2 2 2 5 6]

to

[1 2 2 5 6]

i.e. only leave one third of the repeated elements.

Not sure if this is possible but worth a shot, thanks! :)

[Stephen23]

How do you count "repeats": only contiguous repeats or any repeats throughout the entire vector?

I.e. does this count as two independent groups of 2, or do they all get counted together?:

[1,2,2,2,0,2,2,2,5,6]

[Thomas Gvero]

That would be 2 independent groups.

[Matt J]

Do the repeated elements always occur in fixed, known multiples? How would, a reduction by 3 handle this,

[1 1 1 5 2 2 2 2 6] 

[Thomas Gvero]

As far as I'm aware, in the context i am using it, yes.

Answer 1

[Stephen23]

>> V = [1,2,2,2,2,2,2,5,6];
>> N = 3;
>> X = cumsum([1,diff(V)~=0]);
>> F = @(v) {v(1:ceil(numel(v)/N))};
>> cell2mat(accumarray(X(:),V(:),[],F))
ans =
   1
   2
   2
   5
   6

[Thomas Gvero]

thank you so much!

Answer 2

[Matt J]

Another method, one which avoids for-loops inherent in cell2mat,

V=[1,2,2,2,0,2,2,2,5,5,5,5,5,5,6,6,6];


N=3;

X = cumsum([1,diff(V)~=0]);

[~,idx]=unique([X,inf],'stable');

result=V(  repelem(idx(1:end-1), ceil(diff(idx)/N) )  )

result =

     1     2     0     2     5     5     6