Keep getting error in splitapply

I keep getting this error but I checked my code again and again and can't find anything.

Error using vertcat Dimensions of matrices being concatenated are not consistent.

Error in splitapply>localapply (line 250) finalOut{curVar} = vertcat(funOut{:,curVar});

Error in splitapply (line 133) varargout = localapply(fun,splitData,gdim,nargout);

Could anyone tell me why this is happening?

Jennfier

Answer 1

[Gillian Rosen]

Hi Jennifer, 

I understand that when you try to run your code, you are encountering an error related to the 'splitapply' function. 

If you are using the 'splitapply' function with the 'mean' or 'sum' functions, then you may be encountering this error due to the way 'mean' and 'sum' work differently for matrices vs. vectors. To use 'mean' as an example, it computes the column-wise average for matrices: 

>> mean([1 1; 1 3]) 
ans = 
      1     2 

But it computes the overall average for vectors:

>> mean([2 4]) 
ans = 
      3 

These results cannot be vertically concatenated. Therefore, if you try using 'mean' with 'splitapply', such as:  

>> a = [1 1; 1 3; 2 4]; 
>> b = [1; 1; 2]; 
>> splitapply(@mean, a, b);  % NOT recommended usage

You will encounter the error you described ('Error using vertcat...'). To avoid this error, you can specify that 'mean' should return the mean along rows instead of columns. You can do this by using a slightly different function handle:

>> splitapply(@(x)mean(x, 1), a, b);

[JFz]

Thank you!