Proc Gplot line type option in symbol

[sfo]

I need to give different line option depending upon the variable study in my dataset i.e. if the dataset has study = "study1" then the line=1 and if the data has study="study2" then line=2. Also, I am plotting the graph for various doses. Is there a way I can specify in SAS to change the line option ?

[MikeZdeb]

hi ... here's an example where line color and type varies by the value of the variable sex (there are 46 line patterns, 1 is solid, etc.)

look at the SYMBOL statements ... l=1 is short for line=1

proc sort data=sashelp.class out=class;

by height;

run;

goptions reset=all;

legend1 position=(top inside right) label=none across=1 shape=symbol(.0005,1) value=('FEMALE' 'MALE');

symbol1 v='6c'x f='wingdings' c=red  i=smp66 l=1;

symbol2 v='6c'x f='wingdings' c=blue i=smp66 l=3;

proc gplot data=class;

plot weight*height=sex / legend=legend1;

run;

quit;

ps  lots of info in a SUGI paper (http://www2.sas.com/proceedings/sugi31/239-31.pdf)

[DanH_sas]

Hey,

Take a look at these two blog posts. I deal with this very issue and how you can make this happen using ODS graphics.

Hope this helps!

Dan

http://blogs.sas.com/content/graphicallyspeaking/2012/02/27/roses-are-red-violets-are-blue/

http://blogs.sas.com/content/graphicallyspeaking/2012/03/01/roses-are-red-violets-are-blue-part-2/

[MikeZdeb]

hi ... what do you mean by "line statement"

do you really mean SYMBOL statement ?

also, this ...

plot conc_ngml*Nominal_Time_day=subject_alias;

you are putting 10 lines on one plot, 15 on another =  unreadable yes/no?

[MikeZdeb]

hi ... using your data (too many lines to make any sense ???) ...

libname x 'z:\gplot data.xls';

proc sort data=x."'#LN00453$'"n out=x;

by dose;

run;

goptions reset=all ftext='calibri' htext=2 gunit=pct;

symbol1 i=join v='6c'x f='wingdings' w=4 h=2 color=green;

symbol2 i=join v='6c'x f='wingdings' w=4 h=2 color=blue;

symbol3 i=join v='6c'x f='wingdings' w=4 h=2 color=grey;

symbol4 i=join v='6c'x f='wingdings' w=4 h=2 color=orange;

symbol5 i=join v='6c'x f='wingdings' w=4 h=2 color=red;

symbol6  i=join v='6e'x f='wingdings' w=4 h=2 color=green;

symbol7  i=join v='6e'x f='wingdings' w=4 h=2 color=blue;

symbol8  i=join v='6e'x f='wingdings' w=4 h=2 color=grey;

symbol9  i=join v='6e'x f='wingdings' w=4 h=2 color=orange;

symbol10 i=join v='6e'x f='wingdings' w=4 h=2 color=red;

symbol11 i=join v='ab'x f='wingdings' w=4 h=2 color=green;

symbol12 i=join v='ab'x f='wingdings' w=4 h=2 color=blue;

symbol13 i=join v='ab'x f='wingdings' w=4 h=2 color=grey;

symbol14 i=join v='ab'x f='wingdings' w=4 h=2 color=orange;

symbol15 i=join v='ab'x f='wingdings' w=4 h=2 color=red;

legend1 position=(top inside right) mode=protect across=1 label=none shape=symbol(.0005,2)pct;

axis1 offset=(2,2)pct;

axis2 label=(a=90);

title1 a=90 ls=2;

title2 a=-90 ls=2;

proc gplot data = x;

plot conc_ngml*Nominal__Time=subject__alias / legend=legend1 haxis=axis vaxis=axis2 noframe;

by dose;

format conc_ngml 4.1;

run;

quit;

[sfo]

Thanks for the code. Yes, I know too many lines on one plot will be very messy. But, I need to prepare this plot to see the difference between the two studies.

Is it also possible to get dotted lines (i.e using line=2) option whenever a subject is coming from study 2 and line = 1whenever a subject is coming from study 1?

Your code was really helpful. Thanks again.

[MikeZdeb]

hi ... the attached SAS code produced the two attached plots

I just took out the "by dose" and did two separate plots, changing the symbols from plot1 to plot2 to conform with the number of people in each study

I think it's correct, but you should check the allocation of people to study

[sfo]

Thanks a lot for your help.

So, we have to create two codes for different dose levels. I was hoping to make one code and tweak the symbol option so that it takes line=1 for study1 and line=2 for study2, maybe by using some macro etc. Because I will having more than 2 dose groups in my dataset and I wanted to avoid repeating my code.

[MikeZdeb]

hi ... given the choice of trying to write SAS code that would figure out the SYMBOL statement based on the data versus just doing with I did, I would say that the choice depends on how often you are going to do this

for the two plots, the effort of cutting/pasting code took a LOT less time that writing SAS code that would write SAS code for

or "when to hold 'em, when to fold 'em"