Summation of the quarterly observations

[VinodKavery]

Hi there, I have a SAS Dataset consisting of the Data as mentioned below. These are quarterly figures and do not have an ID column. The number of observations can vary and it need not be 31. I would like to sum the last 4 obs from the bottom of predicted column and skip those 4 & again sum the next last 4 observations.

It is like year2 = 1500 + 1500 +1500 +1500 = 6000 and year1=1000 +1000 +1000 +1000 = 4000.

Once this is done, calculate from Actual column .year0 = 900+ 900+900+900 =3600. Post this calculate the growth percentage numbers i.e., year1_gr = (year1/year0-1)*100 and  year2_gr = (year2/year1-1)*100

Could anyone please help me with the SAS code in doing it?

Obs	Actual	Predicted
1	1,069	1,293
2	1,320	1,274
3	1,284	1,307
4	1,318	1,473
5	1,351	1,241
6	1,371	1,263
7	1,123	1,347
8	1,060	1,357
9	1,366	1,420
10	1,080	1,444
11	1,259	1,393
12	1,467	1,239
13	1,034	1,486
14	1,290	1,114
15	1,367	1,327
16	1,391	1,325
17	1,218	1,353
18	1,364	1,188
19	1,372	1,116
20	900	1,127
21	900	1,383
22	900	1,100
23	900	1,134
24	Missing	1,000
25	Missing	1,000
26	Missing	1,000
27	Missing	1,000
28	Missing	1,500
29	Missing	1,500
30	Missing	1,500
31	Missing	1,500

[Kurt_Bremser · Accepted solution]

data inter;
set have;
no = _n_;
run;

proc sort data=inter;
by descending no;
run;

data want (keep=year2 year1 year0);
array year {3} year2-year0;
do j = 1 to 3;
  year{j} = 0;
  do i = 1 to 4;
    set inter;
    if j ne 3
    then year{j} = year{j} + predicted;
    else year{j} = year{j} + actual;
  end;
end;
output;
stop;
run;

First reverse the order by observation number

Then extract the data in groups of 4, and add up in an array

[VinodKavery]

Thank you sir