topic Re: Opimisation SAS code in SAS Programming

Opimisation SAS code

kelxxx — Thu, 20 Apr 2023 15:37:48 GMT

Hello, my have and want tables are as following

data have;
input id_appel id_demande $ demand_label $20. ;
cards;
1 A-01 Appointement
1 C-09 Contract
2 A-08 Appointement
2 D-10 Devis
2 C-11 Contract
2 C-29 Contract
;
run;
data want;
input id_appel nb_dmd_A nb_dmd_B nb_dmd_C nb_dmd_D ;
cards ;
1 1 0 1 0
2 1 0 2 1
;
run;

This code is work for me

proc sort data= have;
by id_appel demand_label;
quit;

data want;
set have;
by id_appel;

if first.id_appel then do;
nb_dmd_A=0;
nb_dmd_B=0;
nb_dmd_C=0;
nb_dmd_D=0;
end;

retain nb_dmd_: 0;

if demand_label="Appointement" then nb_dmd_A = nb_dmd_A+1;
if demand_label="BBB" then nb_dmd_B = nb_dmd_B+ 1; 
if demand_label="Contract" then nb_dmd_C = nb_dmd_C+1;
if demand_label="Devis" then nb_dmd_D= nb_dmd_D+ 1; 

if last.id_appel;
drop id_demande demand_label;
run;

I am looking to optimise my codes because i have many demand_label different .

Can my codes be optimized ? Can you help me please?

If you have any suggestion i am open for more details ...

Thanks. Regards!!!

Re: Opimisation SAS code

AMSAS — Thu, 20 Apr 2023 16:03:37 GMT

Take a look at PROC MEANS and PROC SUMMARY

proc means data=have n ;
	class id_appel demand_label ;
	var id_appel ;
run ;

Re: Opimisation SAS code

Astounding — Thu, 20 Apr 2023 16:21:53 GMT

Is there any reason you need to use DEMAND_LABEL in your program? Can't you just use the first character in ID_DEMANDE instead? This might get you most of the way there:

proc freq data=have;
   tables id_appel * id_demande / noprint out=counts (keep=id_appel id_demande count);
   format id_demande $1.;
run;

You still need to reformat the counts (including adding zeros where appropriate ... see if the SPARSE option would do that for you), but wouldn't those counts be correct?

Re: Opimisation SAS code

Reeza — Thu, 20 Apr 2023 16:36:07 GMT

Dynamic code is preferable IMO even if slightly longer.

Use the PRELOADFMT and PROC FREQ to generate all combinations for results
PROC TABULATE has the displayed results if that's all you need.
Transpose data from PROC TABULATE into data set, if desired.

I'll leave the renaming of columns up to you.

data have;
input id_appel id_demande $ demand_label $20. ;
cards;
1 A-01 Appointement
1 C-09 Contract
2 A-08 Appointement
2 D-10 Devis
2 C-11 Contract
2 C-29 Contract
;
run;


proc sort data= have;
by id_appel id_demande;
quit;

proc format;
value $ demande_list 
'A' = 'A'
'B' = 'B'
'C' = 'C'
'D' = 'D';
run;

proc tabulate data=have out=long;
class id_appel;
class id_demande / preloadfmt;
format id_demande $demande_list1.;
table id_appel, id_demande*N='' / printmiss misstext='0';
run;

data long;
set long;
if missing(N) then N=0;
run;

proc transpose data=long out=want;
by id_appel;
id id_demande;
var N;
run;

Re: Opimisation SAS code

A_Kh — Thu, 20 Apr 2023 17:51:49 GMT

Like the use of PRELOADFMT option.

Re: Opimisation SAS code

vijaypratap0195 — Fri, 21 Apr 2023 06:28:21 GMT

Hope this will help you-

proc freq data=have;
tables id_appel*demand_label / list nocol norow out=work.tmp(drop=percent) ;
run;

proc sort data=tmp1; by demand_label; run;

data tmp1;
length demand_label $1.;
set tmp;
run;

proc transpose data=tmp1 out=tmp2;
by id_appel;
id demand_label;
var count;
run;