MATLAB Answers

0

Use rowfun to sum multiple columns by group

Asked by Ledger Yu on 15 Oct 2019
Latest activity Commented on by Guillaume
on 15 Oct 2019
I have a table that looks like this:
Data =
date id flag1 flag2
2019-01-01 x1 1 1
2019-01-01 x1 1 0
2019-01-02 x2 0 0
2019-01-02 x2 0 1
...
The line below will sum "flag1" by "date":
rowfun(@sum, Data, 'groupingVariables','date','inputVariables','flag1')
But how do I apply this to both "flag1" and "flag2"? I tried:
rowfun(@sum, Data, 'goupingVariables','date','inputVariables',{'flag1','flag2'})
...and it throws out the following error:
Dimension argument must be a positive integer scalar within indexing range

  0 Comments

Sign in to comment.

Tags

Products


Release

R2018b

1 Answer

Answer by Guillaume
on 15 Oct 2019
 Accepted Answer

rowfun is not the correct function for this. rowfun applies the function by rows and consider the input variables as separate input arguments to the function. Indeed the 2nd argument to sum is the dimension to operate along so you get an error since your 2nd variable does not contain valid dimensions.
You want to apply the same function to the different variables of your table. The function for that is varfun:
varfun(@sum, Data, 'GroupingVariables', 'date', 'InputVariables', {'flag1','flag2'})
However, note that since R2018a we have groupsummary which is even easier to use and allows you to get several statistics at the same time:
groupsummary(Data, 'date', 'sum', {'flag1', 'flag2'})
is the equivalent to the above, but you could also do:
groupsummary(Data, 'date', 'weekly', {'sum', 'mean', 'std'}, {'flag1', 'flag2'})
to get weekly sum, mean and standard deviation at once.

  2 Comments

Rik
on 15 Oct 2019
You can still do something with rowfun, although I'm not sure this is what OP wants to do:
Data=table(...
datetime({'2019-01-01';'2019-01-01';'2019-01-02';'2019-01-02'}),...
{'x1';'x1';'x2';'x2'},...
[1;1;0;0],...
[1;0;0;1],...
'VariableNames',{'date','id','flag1','flag2'});
fun=@(flag1,flag2) sum([flag1;flag2]);
output=rowfun(fun,Data,...
'groupingVariables','date',...
'InputVariables',{'flag1','flag2'})
Oh yes, you can still use rowfun by writing your own function that takes multiple outputs, but the OP question is clearly meant to operate on variables not rows, so varfun is far more appropriate.

Sign in to comment.