About TashaChapman

TashaChapman · ‎03-19-2013

I had a feeling it would be something like that. Thanks for the suggestion! Actually, what I ended up doing is adding the rbreak to the first pass-through PROC Report statement. In the end I got an extra observation in my testrep dataset with the sums I needed. This also allowed me to give the row a header of "Total" in the dataset by changing the "AgeGroup" value for that row. So far it's working like a charm.

TashaChapman · ‎03-19-2013

I'm having some issues with PROC Report and I'm just banging my head against the wall here (it's probably something stupid)... I'm passing my data through PROC Report twice - once to produce some calculations, etc., and the other to create the final report. The issue I'm having is with step two. Here is the data: AgeGroup _C2_ _C3_ change1 pchange1 _C4_ change2 pchange2 _C5_ change3 pchange3 Less than 1 25126 25152 -26 -0.00103 25102 24 0.000955 23710 1416 0.05636 1 to 5 117318 117668 -350 -0.00298 118066 -748 -0.00638 87173 30145 0.25695 6 to 12 141621 141476 145 0.001024 137402 4219 0.029791 93923 47698 0.3368 13 to 18 102062 102078 -16 -0.00016 98468 3594 0.035214 65267 36795 0.36052 And here is my code: ods pdf file="Test.pdf"; proc report nowd data=testrep; column agegroup ("Jan-13" _c2_) ("Dec-12" _c3_ change1 pchange1) ("Jan-12" _c4_ change2 pchange2) ("Healthy Kids Baseline" ("Jun-09" _c5_ change3 pchange3)); define agegroup / display 'Age'; define _c2_ / display 'Current' format=comma12.; define _c3_ / display 'Previous Month' format=comma12.; define change1 / display 'Change' format=comma12.; define pchange1 / display '% Change' format=percentn14.2; define _c4_ / display 'Previous Year' format=comma12.; define change2 / display 'Change' format=comma12.; define pchange2 / display '% Change' format=percentn14.2; define _c5_ / display 'Baseline' format=comma12.; define change3 / display 'Change' format=comma12.; define pchange3 / display '% Change' format=percentn14.2; rbreak after / summarize; run; ods pdf close; For some reason the Rbreak is not producing a total row like I would like. In the PDF, it is coming out completely blank. Anyone know why?

TashaChapman · ‎03-15-2013

I can think of a couple ways... you might consider something with the LAG function (which is similar to the DIF function, except it returns the actual value rather than the difference between the two values). You also might be able to do something with SQL, maybe using a subquery. Hai.kuo is right, though. If you provide an example of what you'd like to see, that might help.

TashaChapman · ‎02-25-2013

Duh! <smacks self on head> Tom is totally right. You're using a character informat and applying it to a numeric variable. The informat that you use in your PUT function should be the same variable type as the incoming source data (in this case numeric). It will still output the new variable as character. What I said was also technically kind of true, but that's just because SAS was being nice and forgiving, not because your code was technically correct.

TashaChapman · ‎02-25-2013

When it says in your log that "Age is uninitialized" what that means is that the variable "Age" does not exist in the dataset. I don't know why the LOG always shows this note in blue, when it seems to me that it should be in big bold red letters, but so be it. It is a problem because any calculations that you attempt on based on "Age" will result to missing (because there is no age...). In the example you posted above, it appears that the problem is because you're writing over the same dataset twice. (Although, as Linlin points out, it's a little hard to tell because your LOG notes reference a different dataset than what the code references...) So ignoring the LOG notes that reference HOMEWRK2.BONE, what I see is: data test; <--- you create a dataset WORK.TEST set homewrk2.test; <--- It is based off of HOMEWRK2.TEST age=int((dov-dob)/365.25); <--- You create a new variable in WORK.TEST called "AGE" run; data test; <--- You again create a dataset WORK.TEST, essentially writing over everything you just did. set homewrk2.test; <--- This is also based off of HOMEWRK2.TEST, ignoring everything you just did. if age < 12 then agegrp=1; <--- SAS freaks out because HOMEWRK2.TEST does not have a variable called "AGE" else if 12 <= age < 20 then agegrp=2; else if age >=20 then agegrp=3; run; <--- The end So instead, try this: data test; set homewrk2.test; age=int((dov-dob)/365.25); if age < 12 then agegrp=1; else if 12 <= age < 20 then agegrp=2; else if age >=20 then agegrp=3; run; <--- The end

TashaChapman · ‎02-25-2013

You're going to get the warning regardless, but you should be able to ignore it. If your PROC Contents shows the variable you want coming out as character, you're doing fine.

TashaChapman · ‎02-22-2013

Wow, Reeza, I didn't know that was available. This is the paper I found: http://www.nesug.org/proceedings/nesug06/po/po09.pdf Pretty neat stuff. However, that seems like an awful lot of work to go through to use PROC Freq. You might want to try using a different procedure, like PROC Report or PROC Tabulate, which lets you customize that kind of thing more easily.

TashaChapman · ‎02-22-2013

You need to have both the LABEL and SUMLABEL options turned on. Try this. It almost gets you there, but there's still an extra row for the grand total, even though the font is white... proc print data=testdata noobs label sumlabel style(grandtotal) = {backgrouncolor=white color=white}; by dummy; var Org Unit N; where Org="Org1"; sum N; label N="Number of x"; label Org="Organization"; label unit="Unit"; label dummy = "Total"; run;

TashaChapman · ‎02-21-2013

Hi Cynthia, I meant basic from a user standpoint, as in "Of course I'd want to be able to do that." I know that doesn't always correspond with easy to implement. Right now I'm not working on a specific report in particular. I'm actually putting together some training documentation, and I was looking for an easy way to group based on computed values, since I thought that might be a question that would come up. Thanks! Tasha

TashaChapman · ‎02-21-2013

That is 100% correct. Formats, even formats applied to a variable in a DATA step and part of the dataset attributes, do not change the underlying values of the variable. So even if it prints as "Caucasian" the unformmatted value should still be 1. This unformmatted value is what you need to use in your new format statement.

TashaChapman · ‎02-21-2013

I'm guessing that it's not, based on some other posts that I've seen, but I'm really, really hoping that I'm wrong. Grouping and sorting based on computed values seems like a basic functionality that is oddly missing from Proc Report.

TashaChapman · ‎02-21-2013

Whether the format is character or numeric is based on the source data, not the output value. If the incoming source data is character, than the format name should begin with a $. If the incoming source data is numeric, then the format name should not begin with a $, but the values to the left of the = should be actual numbers. What you have listed in your original post is a mix of these two, so you can see why people are confused. Also, creating a format with PROC Format, does not actually associate the format with any given variable. It merely creates a template that can be used later. So when you say that you "created the numeric value of Race" using a previous PROC Format, that's not actually accurate. So, if the original input data is numeric then you might use this format: proc format; value Race 1= 'White' other ='Other'; However, if the original input data is character then you might use this format: proc format; value $Race Caucasian = 'White' other = 'Other'; Also, be aware that character formats are case sensitive...

TashaChapman · ‎02-21-2013

In PROC Report, is it possible to group using a computed variable? For example, in the PROC below, I'm creating a computed variable (rbcgen) based on the values of RBC and GENDER. Assuming, of course, that I change the other variable types from display to group or analysis, would it be possible to group by rbcgen? Tasha P.S. I know how to do this in a DATA step (and other workarounds). I'm just using a very simple example to ask the question about the capabilities of PROC Report. data bloodRBC; infile datalines; input gender $ rbc; datalines; M 5.5 M 4.8 M 6.2 M 6.5 M 5.9 M 5.2 M 5.9 F 5.9 F 4.5 F 5.2 F 5.5 F 4.3 F 3.9 F 4.5 ; proc report data=bloodRBC nowd; column gender rbc rbcgen; define gender / display; define rbc / display; define rbcgen / computed; compute rbcgen / character length=10; if rbc = . then rbcgen = 'N/A'; else if gender = 'M' and rbc lt 4.7 then rbcgen = 'Low RBC'; else if gender = 'M' and 4.7 le rbc le 6.1 then rbcgen = 'Normal'; else if gender = 'M' and rbc gt 6.1 then rbcgen = 'High RBC'; else if gender = 'F' and rbc lt 4.2 then rbcgen = 'Low RBC'; else if gender = 'F' and 4.2 le rbc le 5.4 then rbcgen = 'Normal'; else if gender = 'F' and rbc gt 5.4 then rbcgen = 'High RBC'; endcomp; run;

Online Status	Offline
Date Last Visited	‎09-01-2015 07:12 AM

Re: PROC Report - RBREAK & Total Row

PROC Report - RBREAK & Total Row

Re: Combining Readmit with Original Visit Data

Re: Why this warning in numeric to charactoer conversion?

Re: Create new age variable then create agegroups

Re: Why this warning in numeric to charactoer conversion?

Re: proc freq format

Re: SUM on proc print

Re: PROC Report - Grouping by Computed Variable

Re: Proc Format to create user defined formats

Re: PROC Report - RBREAK & Total Row

Re: Why this warning in numeric to charactoer conversion?

Re: PROC Report - RBREAK & Total Row

PROC Report - RBREAK & Total Row

Re: Combining Readmit with Original Visit Data

Re: Why this warning in numeric to charactoer conversion?

Re: Create new age variable then create agegroups

Re: Why this warning in numeric to charactoer conversion?

Re: proc freq format

Re: SUM on proc print

Re: PROC Report - Grouping by Computed Variable

Re: Proc Format to create user defined formats

Re: PROC Report - Grouping by Computed Variable

Re: Proc Format to create user defined formats

PROC Report - Grouping by Computed Variable