topic Re: How to avoid plots overlaying and make a visible differentiation in plots in SAS Enterprise Guide

How to avoid plots overlaying and make a visible differentiation in plots

imanojkumar1 — Fri, 25 Nov 2016 14:41:49 GMT

Using the attached data, I am plotting the X, Y values w.r.t. Dates. I am using the following code:

proc sgplot  data=mydata;
series x=x y=y / group=Date break;
yaxis grid;
run;

And this is what plot looks like:

Now question is...

in the above resulting plot I can not identify the 4 plots clearly.

Therefore, I want to do this:

Between plots, can we add some common no. say something like 6 so that there occurs some visible difference.
Thick lines with different colors (so that visually we can identify each)

Re: How to avoid plots overlaying and make a visible differentiation in plots

RW9 — Fri, 25 Nov 2016 15:12:58 GMT

Hi,

Please post test data as a datastep in the body of the post, attached xlsx files are a security risk. With regards to your questions:

What you talk about is jittering. Here are some examples:

http://blogs.sas.com/content/graphicallyspeaking/2013/07/10/make-better-graphs-with-sas-9-4/

http://blogs.sas.com/content/iml/2011/07/05/jittering-to-prevent-overplotting-in-statistical-graphics.html

You could also manipulate your data, i.e. each group gets +6 added to the result, more than the previous group. Do this in a datastep before the sgplot.

To set the size of the lines, you use

lineattrs=(thickness=X)

Where x is the width.

Re: How to avoid plots overlaying and make a visible differentiation in plots

imanojkumar1 — Fri, 25 Nov 2016 15:37:03 GMT

Hi RW9,
Thanks for raising the concerns. I can assure this file is virus checked before uploading.
Also, this data has some 5000 odds rows, so data step might not be a good option on this forum. And that's the reason why I had to upload the xlsx file (which is exported from SAS EG I work on).

Re: How to avoid plots overlaying and make a visible differentiation in plots

RW9 — Fri, 25 Nov 2016 15:57:58 GMT

Its not concerns, its statutory company policy, and security sense for anyone. If you need to post test data, do so as a datastep (to show structure) and only a few of the obsevations necessary to illustrate the problem. You can do this by following this post:

https://communities.sas.com/t5/SAS-Communities-Library/How-to-create-a-data-step-version-of-your-data-AKA-generate/ta-p/258712

Re: How to avoid plots overlaying and make a visible differentiation in plots

Ksharp — Sat, 26 Nov 2016 06:09:23 GMT


proc sgpanel  data=mydata;
panelby date;
series x=x y=y /  break;
run;

Re: How to avoid plots overlaying and make a visible differentiation in plots

imanojkumar1 — Mon, 28 Nov 2016 08:17:29 GMT

Hi KSharp, Thanks for your suggestion. I already tried it. Unfortunately, this way the outcome is in 4 panels. But what I want that all the 4 plots in one panel, as in my question above, but where Date=6Jul16, the plot should be shifted up on Y-axis by 6000 in each row.
So where:
Date=6Jul16 and MeaNo=1,
the new_Y= -40075+6000 = -34075
Date=6Jul16 and MeaNo=2,
the new_Y= -39875+6000 = -33875
and so forth.
I am thinking, this way we can see some shift in the plot and difference between two dates (29Jun and 6Jul).

And I tried this..

PROC SQL;
 create table mydata as 
 SELECT *,
 CASE
 WHEN Date="06JUL2016"d THEN (y+6000)
 ELSE y
 END AS Y_Plus_6K
 FROM mydata;
QUIT;

proc sgplot  data=mydata1;
series x=x y=Y_Plus_6K / group=Date transparency=0.6 LINEATTRS=(thickness=2) break;
yaxis grid;
run;

But this looks horrible. May be you could suggest me something better.
Thanks once again.