County Heat Map

[Whitlea]

SAS 9.4

Hello, 

I am attempting to create a county heat map with count and county name included on the map.  I think that my variable county is clashing with maps.counties variable but I cant tell.  These are the errors that are coming up:

Some observations were discarded when charting County. 

By variables are not properly sorted on data set work.mtcontycnt.

the data set work.mtmap may be incomplete. 

Also I was wondering if there is a procedure in SAS that will categorize my count groups for me that makes sense statistically? 

Attached is the code.

Thanks So Much for the Help!!

[Whitlea]

FYI "maps" on Ln 95 wasn't supposed to be there and was removed.

[ballardw · Accepted solution]

---

@Whitlea wrote:

SAS 9.4

Some observations were discarded when charting County. 

---

Not an error. This occurs when using the same data set for Data= and MAP= because the DATA will have multiple observations for each ID value. The DATA= data set is expected to have one observation per Id (geographical region) unless you request a summary statistic to display per region.

---

By variables are not properly sorted on data set work.mtcontycnt.

Any procedure or data set that uses a BY statement requires the variables to be sorted in the order indicated on the BY statement.

Use proc sort to sort your work.mtcontycnt set to match. I am assuming this occurs with your MERGE.

You could use your CNT variable directly, no reason to create additional variables. Just use a different format.

proc format ; 
value cntgrp 
.         = '<30'
 0 -< 30  = '<30' 
30 - 60   = '30-60' 
61 - 91   = '61-91' 
92 - 122  = '92-122' 
123 - 153 = '123-153'
154 - 184 = '154-184'
185 - 210 = '185-210'; 

That way if you decide to change the ranges then you only need to modify the format and not go back through a data step to add a new, or replace the existing, variable.

---

Also I was wondering if there is a procedure in SAS that will categorize my count groups for me that makes sense statistically? 

---

You would have to describe what you mean by "makes sense statistically". 

As in which statistic, what it measures and such.

You might use the Choro option LEVELS to create a number region grouping values instead of your format.

[Whitlea]

Thanks so much for the levels suggestion!

[ballardw]

---

@Whitlea wrote:

Thanks so much for the levels suggestion!

---

While working out which Level you want you may want to turn on the RANGE option as well to show a legend with the values for each color.

I generally don't leave a final version that way as the Range tends to show the actual values and people are uncomfortable with not knowing where a value that isn't in the data might have gone.