It looks like this panel has the following characteristics:
1. There are two panel variables, for a total of 8 cells. The vertical panel variable has 2 classes, which I will call ROW1 and ROW2. The horizontal panel variable has 4 classes, which I will call COL1-COL4.
2. Within each cell, there is a categorical variable with levels CAT1-CAT4 and a group variable with levels GROUP1-GROUP4. There is also a Y variable (the response).
Study the output from the following example. Then maps my variable names to the variables in your data set:
data Have;
call streaminit(1);
do ROW = 'Row1', 'Row2';
do COL = 'Col1', 'Col2', 'Col3', 'Col4';
do CAT = 'Cat1', 'Cat2', 'Cat3', 'Cat4';
do GROUP = 'Group1', 'Group2', 'Group3', 'Group4';
do Rep = 1 to 10;
Y = rand("Normal");
output;
end;
end;
end;
end;
end;
run;
proc sgpanel data=Have;
panelby COL ROW / onepanel layout=lattice columns=4 novarname;
vbox Y / category=CAT group=GROUP;
run;
Hi @Rick_SAS @Quentin thanks for the reply.
In addition, I realized that I need to also overlay the box plot with a scatter plot of Serostatus(which comprises of seropositive and seronegative), I added it but it did not come out in the graph.
Also the rowaxis is 0.1, 0.3 and 1.0, despite the apparent gap between 0.1, 0.3 and 1.0, in the plot the space in the rows remains the same in the earlier picture which does not seem logical, I tried to use list of string values, value=("0.1" "0.3" "1.0"), it simply ordered it based on the data(I'm assuming it did this by default).
I'm also unable to increase the width/size of the graph, which I think is the reason the column(x) axis are slanted.
How can I control the color of each box?
lastly the panelby and boxed, how can I remove the box.
Below is the code I currently have.
I ask too many questions, please bear with me.
PROC SORT DATA=simulated OUT=simu1;
BY Visit Day ID Strain Agegrp serostatus;
RUN;
/*Changing the structure of the dataset from wide to long such that the cytokine's are represented as a single variable as
well their corresponding values are in another column*/
PROC TRANSPOSE DATA=simu1 OUT=simu2(rename=(col1=med_iqr _name_=cytokine));
BY Visit Day ID Strain Agegrp serostatus;
VAR IFNg TNFa IL_2 CD107a;
RUN;
title;
footnote;
proc sgpanel data=simu2;
/* Use the SORT= option to order the panels */
panelby strain agegrp / rows=2 columns=4 layout=lattice COLHEADERPOS=bottom sort=ascending uniscale=column novarname START=BOTTOMLEFT NOBORDER;
scatter x=serostatus y=med_iqr / jitter filledoutlinedmarkers
markerattrs=(symbol=circlenotfilled size=11)
markerfillattrs=(color=cxfdae6b)
markeroutlineattrs=(color=black)
transparency=0.5;
vbox med_iqr / category=visit group=cytokine fill BOXWIDTH=0.8
lineattrs=(color=black thickness=2)
whiskerattrs=(color=black thickness
medianattrs=(color=black thickness=2);
rowaxis values=(0.1,0.3,1.0)grid display=(nolabel);
colaxis values=("Baseline" "Day 64" "Day 120" "Day 360")display=(nolabel);
run;
the@ cytokine was overridden... I want both legend for cytokine and serostatus to appear at the bottom of the graph.Edit: Updated the code to have SCATTER after VBOX so that scatterplot will show on top of box plots per @Rick_SAS's comment . I didn't update the photo. And whenever I edit code posted here, I end up with problems with line breaks, so if you see VBOX and SCATTER on one line below, that's not what I intended, but the code will work, because semicolons...
One way to specify the symbols for serostatus is to use an attribute map. You can use the keylegend statement to define what should appear in the legend.
It would be helpful if you would post code to create a small amount of example data with your variable names, along with your code to make a graph.
I took Rick's sample data and added a serostatus variable, then created a plot with the overlaid symbols.
Code like:
data Have;
call streaminit(1);
do ROW = 'Row1', 'Row2';
do COL = 'Col1', 'Col2', 'Col3', 'Col4';
do CAT = 'Cat1', 'Cat2', 'Cat3', 'Cat4';
do GROUP = 'Group1', 'Group2', 'Group3', 'Group4';
do Serostatus="Neg","Pos";
do Rep = 1 to 10;
Y = rand("Normal");
output;
end;
end;
end;
end;
end;
end;
run;
data myattr ;
ID="SeroStatus" ;
Value='Neg' ;
MarkerSymbol='Circle' ;
output ;
Value='Pos' ;
MarkerSymbol='Square' ;
output ;
run ;
proc sgpanel data=Have dattrmap=myattr noautolegend;
panelby COL ROW / onepanel layout=lattice columns=4 novarname;
vbox Y / category=CAT group=GROUP name="vbox";
scatter x=cat y=y /group=serostatus ATTRID=SeroStatus jitter name="scatter";
keylegend "scatter"/ title="Serostatus";
keylegend "vbox"/ title="Strain";
run;I agree with Rick that the overlaid scatterplot makes it a bit messy.
Learn how use the CAT functions in SAS to join values from multiple variables into a single value.
Find more tutorials on the SAS Users YouTube channel.
Click image to register for webinar
Classroom Training Available!
Select SAS Training centers are offering in-person courses. View upcoming courses for:
-
1718 views
-
7 likes
-
-
