Dynamic sgplot labels

[ggunaward]

Hi so I'm trying to set up dynamic x axis labels for an sgplot plot, but can't figure out how to do it. I did the calculations  in proc iml and stored them:

/* Calculate skewness and kurtosis for each sample */

proc iml;
use CombinedSamples;
read all;

kurtosis_Normal = kurtosis(Normal);
kurtosis_Cauchy = kurtosis(Cauchy);
kurtosis_Weibull = kurtosis(Weibull);
kurtosis_Laplace = kurtosis(Laplace);
skewness_Normal = skewness(Normal);
skewness_Cauchy = skewness(Cauchy);
skewness_Weibull = skewness(Weibull);
skewness_Laplace = skewness(Laplace);

print "Results";

print kurtosis_Normal kurtosis_Cauchy kurtosis_Weibull kurtosis_Laplace;
print skewness_Normal skewness_Cauchy skewness_Weibull skewness_Laplace;
store _all_ module=_all_;
quit;

Not sure how to add it to the axis of my graph:

/* Set Up for Part C */
proc rank data=CombinedSamples normal=blom out=CombinedSamplesQuant;
var Normal Cauchy Weibull Laplace;
ranks Normal_Quant Cauchy_Quant Weibull_Quant Laplace_Quant;
run;

/* Create Combined QQPlots */
title 'Q-Q Plot for Normal, Cauchy, Weibull, and Laplace Distributions';
proc sgplot data=CombinedSamplesQuant;
scatter x=Normal_Quant y=Normal;
scatter x=Cauchy_Quant y=Cauchy;
scatter x=Weibull_Quant y=Weibull;
scatter x=Laplace_Quant y=Laplace;
xaxis label="Quantiles";
yaxis min=-500 max=500;
run;

Do I need to instead do the calculations in a macro or somehow within sgplot?

[PaigeMiller]

If you want to overlay scatter plots into one big PROC SGPLOT output, you can have a legend created that indicates which color scatterplot belongs to which data. For example, see here: https://documentation.sas.com/doc/en/pgmsascdc/9.4_3.4/grstatproc/n1vkttjoy99wkwn1iyy9leor7xg2.htm but in this case, the values of the x-axis variable are the same in each call to SERIES (same would be true for SCATTER). Your code doesn't show that, but I think you can rearrange your data such that the x-axis is a variable named QUANTILE for each plot.

Do I need to instead do the calculations in a macro or somehow within sgplot?

No.

[ggunaward]

I think you're misunderstanding. I'm not trying to change the plot at all. I'm trying to take the calculations from my proc iml statement, for example kurtosis_Normal and kurtosis_Cauchy, and replace "Quantiles" in xaxix label with them. Is that possible?

[PaigeMiller]

---

@ggunaward wrote:
I think you're misunderstanding. I'm not trying to change the plot at all. I'm trying to take the calculations from my proc iml statement, for example kurtosis_Normal and kurtosis_Cauchy, and replace "Quantiles" in xaxix label with them. Is that possible?

---

Can you show us (even if it is a hand-drawing) what you want?

[ggunaward]

Okay so essentially for that last part:

/* Create Combined QQPlots */
title 'Q-Q Plot for Normal, Cauchy, Weibull, and Laplace Distributions';
proc sgplot data=CombinedSamplesQuant;
scatter x=Normal_Quant y=Normal;
scatter x=Cauchy_Quant y=Cauchy;
scatter x=Weibull_Quant y=Weibull;
scatter x=Laplace_Quant y=Laplace;
xaxis label="Quantiles";
yaxis min=-500 max=500;
run;

I want to replace "xaxis label="Quantiles";"

with "xaxis label="Normal Distribution Kurtosis:" #Kurtosis Variable# "Skewness:" #Skewness Variable#;"

[PaigeMiller]

Do you want four separate plots, one for Normal, one for Cauchy, one for Weibull, one for Laplace?

[ggunaward]

No, just one plot. The plot that I have is actually fine. I'm purposely plotting each scatter plot on the same plot. The problem I'm having is that the data is randomly generated from a distribution, and so I want a label that shows in plain text, the kurtosis and skewness for each distribution whenever the sas page is run. Essentially I'm trying to figure out how to put a calculation into a label, either via a variable or by doing the calculation on the dataset right then and there. Is that possible?

[jimbarbour]

I'm not sure it this is completely applicable, but Rick Wicklin wrote a post a few months ago (February 2021) on customizations.  He talks about different means of doing this including simple things like #ByVar and #ByVal (which I don't think will help us here) but also more complex things using SGPANEL and Macros for customizing things.  I'm not sure if it will be helpful, but I thought it worth mentioning:

https://blogs.sas.com/content/iml/2021/02/22/byvar-byval-keywords-sas-titles.html

Jim

[ggunaward · Accepted solution]

Okay thanks everyone I think I see how to do it now. Within my proc iml statement, I need to put my variables into macro variables like in here:

https://stackoverflow.com/questions/25042043/using-a-vector-generated-in-sas-iml-as-a-macro-variable

and then call the macro variable in one of my sgplot labels like in here:

https://blogs.sas.com/content/iml/2021/02/22/byvar-byval-keywords-sas-titles.html

Does that seem like it would work?

[jimbarbour]

Sounds like it should work.

Jim

[ggunaward]

It worked! Thanks! Also had to reference this for the right quotations for the strings:

https://support.sas.com/documentation/cdl/en/mcrolref/61885/HTML/default/viewer.htm#a001071889.htm