Error in matlab stratified cross validation partitioning "CVPARTITION can have at most one optional argument"

Question

I am trying to perform stratified cross validation as the data is highly imbalanced. The output labels of the data are in the matrix predictionMatrix. It is a 832*1 dimensional matrix with values 0/1. For cross validation, I am using the function cvpartition, but it is generating the error:

CVPARTITION can have at most one optional argument

The code is:

c = cvpartition(predictionMatrix,'KFold',5,'Stratify',true);

Answer 1

You must be using an older version of MATLAB. For example, in MATLAB release R2016a, the cvpartition command takes only one pair of optional arguments as in

c = cvpartition(predictionMatrix,'KFold',5)

and the option ...'Stratify',true is not available at all. So you would get the same error as yours:

I = randi(100,832,1);
predictionMatrix = (I>50);
c = cvpartition(predictionMatrix,'KFold',5,'Stratify',true);

  Error using cvpartition (line 130)
  CVPARTITION can have at most one optional argument

However, in MATLAB release R2018a and beyond, the same code works just fine:

I = randi(100,832,1);
predictionMatrix = (I>50);
c = cvpartition(predictionMatrix,'KFold',5,'Stratify',true)

c = 

K-fold cross validation partition
   NumObservations: 832
       NumTestSets: 5
         TrainSize: 666  665  665  666  666
          TestSize: 166  167  167  166  166