Challenges in SGPanel and hetmapparam

PaalNavestad — Fri, 16 Mar 2018 12:53:22 GMT

Hi I'm working with SGPanel and heatmapparam. I get some unexcpected behaviour and challenges in controlling axis. Below is what I get. Data and code at the bottom.

The y or row axis behaves strangely. Trying to recreate I created some dummy data

Data forSGpanel_problem;
	call streaminit(1234);
	do date='01feb2010'd to '31dec2017'd by 280;
		Panel='pan1';
		x=1;
		number=rand('POISSON',3);
		output;
		Panel='pan1';
		x=2;
		number=rand('POISSON',3.5);
		output;
		
		panel='pan2';
		x=1;
		number=rand('POISSON',4);
		output;
		panel='pan2';
		x=2;
		number=rand('POISSON',4.5);
		output;
	end;
	format date monyy5.;
run;
data forsgpanel_problem;
	set forsgpanel_problem;
	date_txt=put(date,monyy5.);
run;

ods graphics on  / width=900 px height=600 px imagemap;

proc sgpanel data=forSGpanel_problem;	
	where date ge '01jan2015'd;
	panelby panel;
	heatmapparm x=x y=date colorresponse=number/outline;
	text x=x y=date text=date_txt;
*	rowaxis interval=month;
run;

Although the gist of the data being reasonable similar I get a really different graph. I do not get the breaks in the y axis and the boxes are placed in a seemingly erronous fashion.

Does anybody have an idea why this behaves like this. In the work I'm doing at the moment I have to solve this using text on the Y-axis that works well. However I foresee having to use this with more numeric cases further down the line and would like to know the reason and how it can be overcome using the right settings on the axis if possible?

data test;
  infile datalines dsd truncover;
  input ch_ind:32. AIgeo_adp_cl:32. PCT_ROW:32. mnth_txt:$14. mnth:MONYY5. mnth_plt:$12.;
datalines4;
-1,1,8.9009525761,JUL15,JUL15,L08-09-JUL15
0,1,5.1881971043,JUL15,JUL15,L08-09-JUL15
1,1,0.4662548083,JUL15,JUL15,L08-09-JUL15
-1,2,63.341186111,JUL15,JUL15,L08-09-JUL15
0,2,8.4343303867,JUL15,JUL15,L08-09-JUL15
1,2,0,JUL15,JUL15,L08-09-JUL15
-1,3,8.1122605756,JUL15,JUL15,L08-09-JUL15
0,3,3.5235709713,JUL15,JUL15,L08-09-JUL15
1,3,0.3030656254,JUL15,JUL15,L08-09-JUL15
-1,4,0,JUL15,JUL15,L08-09-JUL15
0,4,0,JUL15,JUL15,L08-09-JUL15
1,4,98.379764541,JUL15,JUL15,L08-09-JUL15
-1,5,5.5925432756,JUL15,JUL15,L08-09-JUL15
0,5,5.1320356288,JUL15,JUL15,L08-09-JUL15
1,5,0.3613474764,JUL15,JUL15,L08-09-JUL15
-1,6,11.973778552,JUL15,JUL15,L08-09-JUL15
0,6,5.6161475474,JUL15,JUL15,L08-09-JUL15
1,6,0.4429420678,JUL15,JUL15,L08-09-JUL15
-1,7,2.0383078972,JUL15,JUL15,L08-09-JUL15
0,7,8.5657482393,JUL15,JUL15,L08-09-JUL15
1,7,0.0466254808,JUL15,JUL15,L08-09-JUL15
-1,8,0.040971013,JUL15,JUL15,L08-09-JUL15
0,8,63.539970122,JUL15,JUL15,L08-09-JUL15
1,8,0,JUL15,JUL15,L08-09-JUL15
-1,1,4.186728072,APR16,APR16,L09-10-APR16
0,1,5.252798498,APR16,APR16,L09-10-APR16
1,1,3.519727505,APR16,APR16,L09-10-APR16
-1,2,5.23341009,APR16,APR16,L09-10-APR16
0,2,13.141080639,APR16,APR16,L09-10-APR16
1,2,12.035197275,APR16,APR16,L09-10-APR16
-1,3,41.720745238,APR16,APR16,L09-10-APR16
0,3,1.9783791422,APR16,APR16,L09-10-APR16
1,3,0.0567697985,APR16,APR16,L09-10-APR16
-1,4,5.1287418882,APR16,APR16,L09-10-APR16
0,4,8.0336324454,APR16,APR16,L09-10-APR16
1,4,6.6988362191,APR16,APR16,L09-10-APR16
-1,5,9.8806782499,APR16,APR16,L09-10-APR16
0,5,4.6441440987,APR16,APR16,L09-10-APR16
1,5,2.0720976441,APR16,APR16,L09-10-APR16
-1,6,3.5377852208,APR16,APR16,L09-10-APR16
0,6,5.9411936894,APR16,APR16,L09-10-APR16
1,6,4.3145046835,APR16,APR16,L09-10-APR16
-1,7,0.0628009211,APR16,APR16,L09-10-APR16
0,7,5.4365051327,APR16,APR16,L09-10-APR16
1,7,69.25915413,APR16,APR16,L09-10-APR16
-1,8,30.24911032,APR16,APR16,L09-10-APR16
0,8,55.572266354,APR16,APR16,L09-10-APR16
1,8,2.0437127448,APR16,APR16,L09-10-APR16
-1,1,5.5873925501,OCT16,OCT16,L10-11-OCT16
0,1,5.2175836996,OCT16,OCT16,L10-11-OCT16
1,1,3.2598920863,OCT16,OCT16,L10-11-OCT16
-1,2,4.7564469914,OCT16,OCT16,L10-11-OCT16
0,2,13.536753392,OCT16,OCT16,L10-11-OCT16
1,2,1.506294964,OCT16,OCT16,L10-11-OCT16
-1,3,0,OCT16,OCT16,L10-11-OCT16
0,3,0,OCT16,OCT16,L10-11-OCT16
1,3,88.916366906,OCT16,OCT16,L10-11-OCT16
-1,4,22.836676218,OCT16,OCT16,L10-11-OCT16
0,4,7.6031096316,OCT16,OCT16,L10-11-OCT16
1,4,1.8660071942,OCT16,OCT16,L10-11-OCT16
-1,5,4.3266475645,OCT16,OCT16,L10-11-OCT16
0,5,4.9993462935,OCT16,OCT16,L10-11-OCT16
1,5,0.5395683453,OCT16,OCT16,L10-11-OCT16
-1,6,5.5873925501,OCT16,OCT16,L10-11-OCT16
0,6,5.9145354158,OCT16,OCT16,L10-11-OCT16
1,6,2.9451438849,OCT16,OCT16,L10-11-OCT16
-1,7,53.782234957,OCT16,OCT16,L10-11-OCT16
0,7,5.9789003651,OCT16,OCT16,L10-11-OCT16
1,7,0.1573741007,OCT16,OCT16,L10-11-OCT16
-1,8,3.1232091691,OCT16,OCT16,L10-11-OCT16
0,8,56.749771203,OCT16,OCT16,L10-11-OCT16
1,8,0.809352518,OCT16,OCT16,L10-11-OCT16
-1,1,5.4568527919,MAY17,MAY17,L11-12-MAY17
0,1,5.0484297036,MAY17,MAY17,L11-12-MAY17
1,1,8.544869503,MAY17,MAY17,L11-12-MAY17
-1,2,5.7106598985,MAY17,MAY17,L11-12-MAY17
0,2,12.231679875,MAY17,MAY17,L11-12-MAY17
1,2,37.754737218,MAY17,MAY17,L11-12-MAY17
-1,3,22.165820643,MAY17,MAY17,L11-12-MAY17
0,3,3.3411603561,MAY17,MAY17,L11-12-MAY17
1,3,0.572041473,MAY17,MAY17,L11-12-MAY17
-1,4,37.182741117,MAY17,MAY17,L11-12-MAY17
0,4,7.3652284512,MAY17,MAY17,L11-12-MAY17
1,4,1.1798355381,MAY17,MAY17,L11-12-MAY17
-1,5,4.1878172589,MAY17,MAY17,L11-12-MAY17
0,5,4.531846199,MAY17,MAY17,L11-12-MAY17
1,5,14.837325706,MAY17,MAY17,L11-12-MAY17
-1,6,5.0338409475,MAY17,MAY17,L11-12-MAY17
0,6,5.6403483025,MAY17,MAY17,L11-12-MAY17
1,6,11.548087236,MAY17,MAY17,L11-12-MAY17
-1,7,8.6717428088,MAY17,MAY17,L11-12-MAY17
0,7,7.0100772919,MAY17,MAY17,L11-12-MAY17
1,7,16.410439757,MAY17,MAY17,L11-12-MAY17
-1,8,11.590524535,MAY17,MAY17,L11-12-MAY17
0,8,54.831229821,MAY17,MAY17,L11-12-MAY17
1,8,9.1526635681,MAY17,MAY17,L11-12-MAY17
-1,1,0,OCT17,OCT17,L12-13-OCT17
0,1,5.1485037859,OCT17,OCT17,L12-13-OCT17
1,1,0,OCT17,OCT17,L12-13-OCT17
-1,2,0,OCT17,OCT17,L12-13-OCT17
0,2,12.752977992,OCT17,OCT17,L12-13-OCT17
1,2,0,OCT17,OCT17,L12-13-OCT17
-1,3,0,OCT17,OCT17,L12-13-OCT17
0,3,3.6834899554,OCT17,OCT17,L12-13-OCT17
1,3,0,OCT17,OCT17,L12-13-OCT17
-1,4,0,OCT17,OCT17,L12-13-OCT17
0,4,7.8605955053,OCT17,OCT17,L12-13-OCT17
1,4,0,OCT17,OCT17,L12-13-OCT17
-1,5,0,OCT17,OCT17,L12-13-OCT17
0,5,4.7927280178,OCT17,OCT17,L12-13-OCT17
1,5,0,OCT17,OCT17,L12-13-OCT17
-1,6,0,OCT17,OCT17,L12-13-OCT17
0,6,5.780890557,OCT17,OCT17,L12-13-OCT17
1,6,0,OCT17,OCT17,L12-13-OCT17
-1,7,0,OCT17,OCT17,L12-13-OCT17
0,7,7.2915405463,OCT17,OCT17,L12-13-OCT17
1,7,0,OCT17,OCT17,L12-13-OCT17
-1,8,0,OCT17,OCT17,L12-13-OCT17
0,8,52.68927364,OCT17,OCT17,L12-13-OCT17
1,8,0,OCT17,OCT17,L12-13-OCT17
;;;;
run;

proc sgpanel data=test;
	format mnth monyy5.;
	panelby ch_ind / columns=3;
	heatmapparm x=aigeo_adp_cl y=mnth colorresponse=pct_row/ outline ;
run;

Re: Challenges in SGPanel and hetmapparam

ballardw — Fri, 16 Mar 2018 15:53:01 GMT

First a comment on using:

data forsgpanel_problem;
set forsgpanel_problem;
date_txt=put(date,monyy5.);
run;

Habitual use of that structure will eventually cause a problem with hard to trace down data errors. In your example the variable could be added as

do date='01feb2010'd to '31dec2017'd by 280;
      date_txt=put(date,monyy5.);

Second it is not obvious which graph you want, which goes with which code, exactly what is wrong,and you say "The y or row axis behaves strangely." without describing what "strangely" may mean.

If the issue is that gaps appear in the values that is a function of your MNTH values, they are not evenly spaced: Jul15, Apr16, Oct16, May17 and Oct17 which becomes a bit more obvious if you add

   rowaxis values=('01JUL15'd '01Apr16'd '01Oct16'd '01May17'd '01Oct17'd);

to the second code.

Re: Challenges in SGPanel and hetmapparam

PaalNavestad — Sun, 18 Mar 2018 08:57:51 GMT

Hi ballardw, sry for not making myself clear. The challenge is that to similar datasets give different results, none of them being right.

Controllable solution is to use a character variable. As you can see there is one on the second code snippet. using this give a nice result. However you have to either add a variable that sorts right, For dates using yyyymm or a prefix and monyy5. format. In the current case I'm working this is acceptable. But future cases it might not be. Thanks for your interest.

topic Challenges in SGPanel and hetmapparam in Graphics Programming

Challenges in SGPanel and hetmapparam

Re: Challenges in SGPanel and hetmapparam

Re: Challenges in SGPanel and hetmapparam