cumulative sum by date

[Andres_Fuentes1]

Dear, I hope you are very well, my question is the following: I have 3 columns with numerical variables and I must accumulate them from the oldest to the newest as I show in the following table of how the data is and how the final table should be.

DATE	CATEG_A	CATEG_B	CATEG_C
16-01-2022	1	5	0
15-01-2022	8	1	3
14-01-2022	4	0	1
13-01-2022	1	1	0
12-01-2022	1	0	0

Results.

DATE	CATEG_A	CUM_A	CATEG_B	CUM_B	CATEG_C	CUM_C
16-01-2022	1	15	5	7	0	4
15-01-2022	8	14	1	2	3	4
14-01-2022	4	6	0	1	1	1
13-01-2022	1	2	1	1	0	0
12-01-2022	1	1	0	0	0	0

thanks for your time

[PaigeMiller]

Please explain how in the original data CATEG_C is always zero but in the output CUM_C is sometimes not zero. And lots of other discrepancies between the input and output also exist and require explanation. Thanks.

[Andres_Fuentes1]

sorry, I changed the table for a smaller one for the example, I will renew it.

[PaigeMiller · Accepted solution]

If you have PROC EXPAND, this will do cumulative sums as follows:

proc sort data=have;
    by date;
run;
proc expand data=have out=want;
    convert categ_a=cum_a categ_b=cum_b categ_c=cum_c/transformout=(cusum);
run;

Otherwise you would have to do this in a DATA step.

proc sort data=have;
    by date;
run;
data want;
    set have;
    cum_a+categ_a;
    cum_b+categ_b;
    cum_c+categ_c;
run;