topic coordination of dataset with splot procedure in Graphics Programming

coordination of dataset with splot procedure

MaryA_Marion — Thu, 18 Nov 2021 04:42:38 GMT

SAS STUDIO Release: 3.8 (Enterprise Edition)

Is there a description of what form dataset should be in for sgplot and other s series plots? I have the following:

proc sgplot data=profile;
series x=brand y=response / group=subject;
run;

I want a very simple profile plot of lifetime * brand for each subject connecting lines between brands for each subject.

Is there a clue in this code as to what input form to use? I usually use longform. I am finding all sorts of graphs out there with code but coordinating the data set with the procedure is another matter.

DATA profile;
INPUT brand $ response;
subject=_n_;
DATALINES;
A  73
A  64
A  67 
B  81 
B  77
;

Re: coordination of dataset with splot procedure

sbxkoenk — Thu, 18 Nov 2021 12:31:21 GMT

Post moved by me to Graphics board (coming from Visual Analytics board).

Koen

Re: coordination of dataset with splot procedure

sbxkoenk — Thu, 18 Nov 2021 12:35:34 GMT

Hello,

With series statement in proc sgplot an a unique subject (subject only occurring once) you do not see anything as there's no marker by default and there's nothing to connect.

I think you want something like the below :

DATA profile;
INPUT brand $ response subject $;
*subject=_n_;
DATALINES;
A  73 1
A  64 2
A  67 3
B  81 1
B  77 2
;
run;

proc sgplot data=profile;
series  x=brand y=response / group=subject;
scatter x=brand y=response / group=subject;
run;
/* end of program */

Koen

Re: coordination of dataset with splot procedure

MaryA_Marion — Thu, 18 Nov 2021 14:11:36 GMT

Solution found. I was trying to enter subject numbers easily. They were not correct and all subsequent graphs were incorrect. I went back to the old technique of using scatter and series statements.

DATA profile;
INPUT subject brand $ response;
DATALINES;
1 A  73
2 A  64
3 A  67 
1 B  81 
2 B  77
;
proc sgplot data=profile; 
scatter x=brand y=response / group=subject; 
series x=brand y=response / group=subject; 
run;