creating heat map of correlation matrix

[Solvej]

Dear all

I want to make a heat map of two discrete variables (there are 12 Groups in each). Do you have an suggestions on how to do this?

I use base sas and would just like to Group in positive and negatively correlations. Potentially also strong and less strongly correlations. I have tried to find a tutorial but have not been able to find one with two discrete variables.

Kind regards

Solvej

My sas two by two table looks like this (not real data)

Frequency
Percent
Row Pct
Col Pct	A	B	C	D	E	F	G	H	I	J	K	L	Total
A	10	26	7	9	14	119	140	16	63	1	29	30	464
0.22	0.56	0.15	0.19	0.3	2.57	3.03	0.35	1.36	0.02	0.63	0.65	10.03
2.16	5.6	1.51	1.94	3.02	25.65	30.17	3.45	13.58	0.22	6.25	6.47
30.3	7.14	13.46	7.89	9.72	12.91	8.97	13.56	9.81	7.69	8.92	8.9
B	5	140	1	4	3	25	12	1	20	1	11	29	252
0.11	3.03	0.02	0.09	0.06	0.54	0.26	0.02	0.43	0.02	0.24	0.63	5.45
1.98	55.56	0.4	1.59	1.19	9.92	4.76	0.4	7.94	0.4	4.37	11.51
15.15	38.46	1.92	3.51	2.08	2.71	0.77	0.85	3.12	7.69	3.38	8.61
C	0	5	26	2	7	4	2	3	2	1	2	4	58
0	0.11	0.56	0.04	0.15	0.09	0.04	0.06	0.04	0.02	0.04	0.09	1.25
0	8.62	44.83	3.45	12.07	6.9	3.45	5.17	3.45	1.72	3.45	6.9
0	1.37	50	1.75	4.86	0.43	0.13	2.54	0.31	7.69	0.62	1.19
D	0	1	0	61	5	10	3	0	6	1	3	2	92
0	0.02	0	1.32	0.11	0.22	0.06	0	0.13	0.02	0.06	0.04	1.99
0	1.09	0	66.3	5.43	10.87	3.26	0	6.52	1.09	3.26	2.17
0	0.27	0	53.51	3.47	1.08	0.19	0	0.93	7.69	0.92	0.59
E	1	13	1	2	54	10	7	2	1	0	4	0	95
0.02	0.28	0.02	0.04	1.17	0.22	0.15	0.04	0.02	0	0.09	0	2.05
1.05	13.68	1.05	2.11	56.84	10.53	7.37	2.11	1.05	0	4.21	0
3.03	3.57	1.92	1.75	37.5	1.08	0.45	1.69	0.16	0	1.23	0
F	2	75	5	15	19	550	77	8	115	3	93	35	997
0.04	1.62	0.11	0.32	0.41	11.89	1.67	0.17	2.49	0.06	2.01	0.76	21.56
0.2	7.52	0.5	1.5	1.91	55.17	7.72	0.8	11.53	0.3	9.33	3.51
6.06	20.6	9.62	13.16	13.19	59.65	4.94	6.78	17.91	23.08	28.62	10.39
G	6	13	5	6	24	77	1262	10	3	0	3	16	1425
0.13	0.28	0.11	0.13	0.52	1.67	27.29	0.22	0.06	0	0.06	0.35	30.82
0.42	0.91	0.35	0.42	1.68	5.4	88.56	0.7	0.21	0	0.21	1.12
18.18	3.57	9.62	5.26	16.67	8.35	80.9	8.47	0.47	0	0.92	4.75
H	2	6	0	4	9	19	26	70	10	0	6	2	154
0.04	0.13	0	0.09	0.19	0.41	0.56	1.51	0.22	0	0.13	0.04	3.33
1.3	3.9	0	2.6	5.84	12.34	16.88	45.45	6.49	0	3.9	1.3
6.06	1.65	0	3.51	6.25	2.06	1.67	59.32	1.56	0	1.85	0.59
I	7	52	4	4	4	64	16	3	382	0	35	29	600
0.15	1.12	0.09	0.09	0.09	1.38	0.35	0.06	8.26	0	0.76	0.63	12.98
1.17	8.67	0.67	0.67	0.67	10.67	2.67	0.5	63.67	0	5.83	4.83
21.21	14.29	7.69	3.51	2.78	6.94	1.03	2.54	59.5	0	10.77	8.61
J	0	3	0	0	0	1	0	0	4	4	0	0	12
0	0.06	0	0	0	0.02	0	0	0.09	0.09	0	0	0.26
0	25	0	0	0	8.33	0	0	33.33	33.33	0	0
0	0.82	0	0	0	0.11	0	0	0.62	30.77	0	0
K	0	16	1	2	2	30	8	5	29	1	133	8	235
0	0.35	0.02	0.04	0.04	0.65	0.17	0.11	0.63	0.02	2.88	0.17	5.08
0	6.81	0.43	0.85	0.85	12.77	3.4	2.13	12.34	0.43	56.6	3.4
0	4.4	1.92	1.75	1.39	3.25	0.51	4.24	4.52	7.69	40.92	2.37
L	0	14	2	5	3	13	7	0	7	1	6	182	240
0	0.3	0.04	0.11	0.06	0.28	0.15	0	0.15	0.02	0.13	3.94	5.19
0	5.83	0.83	2.08	1.25	5.42	2.92	0	2.92	0.42	2.5	75.83
0	3.85	3.85	4.39	2.08	1.41	0.45	0	1.09	7.69	1.85	54.01
Total	33	364	52	114	144	922	1560	118	642	13	325	337	4624
0.71	7.87	1.12	2.47	3.11	19.94	33.74	2.55	13.88	0.28	7.03	7.29	100

[PaigeMiller]

I don't understand your table, it is certainly not a correlation matrix, but to create a colored table of anything, you can use PROC REPORT. See this example: http://www.caloxy.com/papers/69-TUT.pdf

[Ksharp]

@Rick_SAS has already written a couple of blogs about it.

Or try 

proc sgplot data=

heatmap ...........

heatmapparm ..........

[PaigeMiller]

@Ksharp

Does SGPLOT produce colored tables?

[WarrenKuhfeld]

These might be of interest.

https://blogs.sas.com/content/graphicallyspeaking/2017/12/11/displaying-upper-lower-triangle-correla...

support.sas.com/resources/papers/proceedings17/SAS0312-2017.pdf 

[PaigeMiller]

---

@WarrenKuhfeld wrote:

These might be of interest.

https://blogs.sas.com/content/graphicallyspeaking/2017/12/11/displaying-upper-lower-triangle-correla...

support.sas.com/resources/papers/proceedings17/SAS0312-2017.pdf 

---

I was not aware of these. Impressive. 

Still, I would think creating such a colored correlation matrix (or other form of colored table) would somehow get easier over time, perhaps even to the point where it was built in to SAS, so people wouldn't need such complicated code to do this.

[Ksharp]

Sure.

Check WarrenKuhfeld and other's blog Graph Speaking .

[Solvej]

I tried writing this

proc sgplot data=diagnose noautolegend; heatmap x= Hdiaggrup Y= epiHdiaggrup / freq=Count discretex discretey colormodel=TwoColorRamp outline; run;

I get the following error statement
372 proc sgplot data=diagnose noautolegend; heatmapparm x= Hdiaggrup Y= epiHdiaggrup
-----------
180
372! / freq=Count discretex discretey
ERROR 180-322: Statement is not valid or it is used out of proper order.

What can be wrong

[Solvej]

I tried writing this

proc sgplot data=diagnose noautolegend;

heatmap x= Hdiaggrup Y= epiHdiaggrup / freq=Count discretex discretey colormodel=TwoColorRamp outline; run;

I get the following error statement
372 proc sgplot data=diagnose noautolegend; heatmapparm x= Hdiaggrup Y= epiHdiaggrup
-----------
180
372! / freq=Count discretex discretey
ERROR 180-322: Statement is not valid or it is used out of proper order.

What can be wrong?

[Ksharp]

Maybe your SAS version is too old. Your sas at least SAS 9.4m4

Or you could try PROC IML ,check @Rick_SAS 's blog 

http://blogs.sas.com/content/iml/

[Ksharp]

Maybe your SAS version is too old. Your sas at least SAS 9.4m4

Or you could try PROC IML ,check @Rick_SAS 's blog 

http://blogs.sas.com/content/iml/

[DanH_sas]

Given the nature of your data, can you  use a HEATMAPPARM instead of a HEATMAP? HEATMAPPARM was released before HEATMAP. Try something like this:

proc sgplot data=diagnose noautolegend;

heatmapparm x= Hdiaggrup Y= epiHdiaggrup colorrresponse=Count /  discretex discretey colormodel=TwoColorRamp outline;

run;

Hope this helps!

Dan