Summing a variable within an ID based on date less 2 days

[mfowlk]

I'm trying find a sum for var1 and separate sum for var2 for each row based on the sas date column labelled dt and two days prior (which is in the sum_to_date) within each ID.

For example,

ID232445

line 5 sum(var1) = 0 sum(var2) =1

line 6 sum(var1) = 0 sum(var2) =2

line 7 sum(var1) = 0 sum(var2) =3

line 8 sum(var1) = 0 sum(var2) =3 

line 9 sum(var1) = 0 sum(var2) =4

line 10 sum(var1) = 0 sum(var2) =3

line 11 sum(var1) = 0 sum(var2) =1

line 12 sum(var1) =0 sum(var2)=2

continue summing until reaching ID 112344, then begin the summing again.

[Reeza]

Please show expected output.

[Quentin]

Can you post this example data as a CARDS data step, for the data you HAVE, along with the data you WANT to create, and also the code you have tried?  Are you familiar with BY-group processing?  Seeing the code you have tried will help others help you. 

I don't really understand how DT is involved in your logic.  Seeing the data you HAVE as input and the data you WANT as output will help.

[mfowlk]

Quentin, 

Thank you! The date that I have as input is in the picture of the spreadsheet. I will create what I want in another spreadsheet.

[Reeza]

Rather than spreadsheet please post your data as text into the forum directly or better yet, as a data step as I did in my answer. Makes it much easier to respond to you.

[mfowlk]

I'm new to this forum. Is this what you mean by 'posting data as text directly into forum'?

HAVE
ID	var1	var2	dt	sum_to_date
232445	0	1	23145	23143
232445	0	0	23145	23143
232445	0	1	23150	23148
232445	0	0	23151	23149
232445	0	1	23151	23149
232445	0	1	23151	23149
232445	0	0	23152	23150
232445	0	1	23152	23150
232445	0	0	23153	23151
232445	0	1	23157	23155
232445	0	1	23157	23155
232445	0	0	23158	23156
232445	0	0	23159	23157
232445	0	0	23159	23157
232445	0	1	23159	23157
232445	0	0	23160	23158
112344	0	0	23135	23133
112344	0	0	23135	23133
112344	0	1	23135	23133
112344	0	1	23136	23134
112344	0	0	23136	23134
112344	0	1	23137	23135
112344	0	1	23137	23135
WANT
ID	var1	var2	dt	sum_to_date	sum(var1)	sum(var2)
232445	0	1	23145	23143
232445	0	0	23145	23143
232445	0	1	23150	23148	0	1
232445	0	0	23151	23149	0	1
232445	0	1	23151	23149	0	2
232445	0	1	23151	23149	0	3
232445	0	0	23152	23150	0	3
232445	0	1	23152	23150	0	4
232445	0	0	23153	23151	0	3
232445	0	1	23157	23155	0	1
232445	0	1	23157	23155	0	2
232445	0	0	23158	23156	0	2
232445	0	0	23159	23157	0	2
232445	0	0	23159	23157	0	2
232445	0	1	23159	23157	0	3
232445	0	0	23160	23158	0	1
112344	0	0	23135	23133	0	0
112344	0	0	23135	23133	0	0
112344	0	1	23135	23133	0	1
112344	0	1	23136	23134	0	2
112344	0	0	23136	23134	0	2
112344	0	1	23137	23135	0	3
112344	0	1	23137	23135	0	4

[Reeza]

You have duplicate rows with nothing that uniquely identifies a row making this a bit messier. 

This seems to roughly work. Not sure of your logic though as you start the answers at 5 but line 4 should also have an answer as well? Are the end of the intervals included?

data have;
infile cards truncover;
input ID $ var1    var2    dt  sum_to_date;
format dt sum_to_date date9.;
cards;
232445  0   1   23145   23143
232445  0   0   23145   23143
232445  0   1   23150   23148
232445  0   0   23151   23149
232445  0   1   23151   23149
232445  0   1   23151   23149
232445  0   0   23152   23150
232445  0   1   23152   23150
232445  0   0   23153   23151
232445  0   1   23157   23155
232445  0   1   23157   23155
232445  0   0   23158   23156
232445  0   0   23159   23157
232445  0   0   23159   23157
232445  0   1   23159   23157
232445  0   0   23160   23158
112344  0   0   23135   23133
112344  0   0   23135   23133
112344  0   1   23135   23133
112344  0   1   23136   23134
112344  0   0   23136   23134
112344  0   1   23137   23135
112344  0   1   23137   23135
;
run;

data have;
set have;
obs= _n_;
run;

proc sql;
create table want as
select distinct t1.*, sum(t2.var1) as sum_var1, sum(t2.var2) as sum_var2
from have as t1
left join have as t2
on t1.id=t2.id and t2.dt between t1.sum_to_date and t1.dt-1
group by t1.obs
order by t1.obs;
quit;

---

@mfowlk wrote:

I'm trying find a sum for var1 and separate sum for var2 for each row based on the sas date column labelled dt and two days prior (which is in the sum_to_date) within each ID.

For example,

ID232445

line 5 sum(var1) = 0 sum(var2) =1

line 6 sum(var1) = 0 sum(var2) =2

line 7 sum(var1) = 0 sum(var2) =3

line 8 sum(var1) = 0 sum(var2) =3 

line 9 sum(var1) = 0 sum(var2) =4

line 10 sum(var1) = 0 sum(var2) =3

line 11 sum(var1) = 0 sum(var2) =1

line 12 sum(var1) =0 sum(var2)=2

continue summing until reaching ID 112344, then begin the summing again.

---

[mfowlk]

Thank you. Yes, lines 1-4 should have answers as well, but I only included 1 month and the dates early in the month will not have prior dates to sum. For the ouput, I would like to include all the columns that are listed for each observation (ID, var1, var2, dt, sum_to_date) as well as the summation of var1 and var2.

Would the monotonic function help to uniquely identify each row?

Yes, the end of the interval would be include in the sum. So the observation/row would be possibly summing 3 numbers if there is a date that is within 2 days prior to the date on the observation/row.