MATLAB: Splitapply cumsum with non-scalar result

Question

I would like to be able to get the accumulating value of numbers from the first vector

>> t = [1 2 3 4 5; 0 0 2 2 2]'

t =

1 0

2 0

3 2

4 2

5 2

cumsum would work very well by writing cumsum(t(:,1)), however, if I wanted the second column to be a grouping value I haven't been able to work out how to write this. If I wanted the sum per group I could:

>> [G,results] = findgroups(t(:,2))

G =

1

1

2

2

2

results =

0

2

and then

>> splitapply(@(x) sum(x), t(:,1),G)

ans =

3

12

3) the thing is I would like to use cumsum in place of sum.

splitapply(@(x) cumsum(x), inputData,G)

Error using splitapply (line 132)

The function '@(x)cumsum(x)' returned a non-scalar value when applied to the 1st group of data.

To compute nonscalar values for each group, create an anonymous function to return each value in a scalar cell:

@(x){cumsum(x)}

I understand what the problem is, but I don't have enough experience to change the function to cater for the result being a non-scalar. I would be ever so grateful if anyone could help in this respect as I unfortunately haven't got much experience in MATLAB with this type of challenge but can see that it would be very useful to solve.

Thank you and kind regards,

William

Answer 1

The error message tells you exactly what to do:

  splitapply(@(x){cumsum(x)}, inputData, G)

This will output a cell array, here each cell contains the cumsum of each group.

Note that in your sum example (and non-working cumsum) the anonymous function is not needed, you can pass directly a handle to sum or cumsum:

  splitapply(@sum, t(:, 1), G)