topic Re: Count the number of days in SAS Enterprise Guide

Count the number of days

Feksan — Tue, 01 Dec 2020 12:34:22 GMT

I have this data set and I want to get the number of days with a number of deaths equal or higher to 1 by iso_code:

DATA MYLIB_D3.NB_DECES;
SET MYLIB_D3.COVID_DATA;
BY date iso_code;
WHERE new_deaths>=1;
RUN;

PROC FREQ data= mylib_d3.nb_deces ;
TABLE date*pays;
RUN;

Re: Count the number of days

PaigeMiller — Tue, 01 Dec 2020 12:36:30 GMT

As stated in your other thread, do not post data as attachments. Post data following these instructions (and not in any other form).

Re: Count the number of days

utrocketeng — Tue, 01 Dec 2020 13:54:33 GMT

proc sql;
create table MYLIB_D3.NB_DECES as
select date, iso_code, count(*) as NbyDate FROM MYLIB_D3.COVID_DATA WHERE new_deaths>=1 GROUP BY date, iso_code;
quit;

Re: Count the number of days

utrocketeng — Tue, 01 Dec 2020 13:55:38 GMT

i should say, this will give you a count of how many per day where new>1. from that, you can get a count of the number of those days there are.

Re: Count the number of days

mkeintz — Tue, 01 Dec 2020 17:47:50 GMT

I won't bother with the data you posted as an attachment, but this is a good situation for running a proc freq on a subset of the original data, and then running a proc freq and the results of the first proc freq:

proc freq data=myhlib_d3.nb_deces noprint;
  where new_deaths>=1;
  table date*iso_code / out=need (keep=iso_code date);
run;
proc freq data=need;
  tables iso_code /out=want (keep=iso_code count);
run;

Note that the WHERE filter can be applied in the first proc freq, so you don't need to run a preparatory DATA step.

The first proc freq writes an output dataset NEED. It would ordinarily have 4 variables: iso_code date count and percent. But you only need the iso_code and date variables.

Then the second proc freq is applied to dataset need, giving a frequency of each iso_code, i.e. then number of unique date values encountered per iso_code.

This is a good example of using the power of various proc's instead of data step programming.