About PaigeMiller

PaigeMiller · ‎09-18-2012

1. Use proc rank to output not only the rank (or bucket) of each observation, but also the minimum and maximum values of that rank (or bucket)? For instance, rank 0 has a minimum value of 1 and a maximum value of 10 rank 1 has a minimum value of 11 and a maximum value of 19 ... rank n has a minimum value of x and a maximum value of y What happens when the item in dataset 2 has a value of 10.5? More generally, the ranks/buckets created from data set one won't come out so nice and clean as your example, it might be more along the lines of rank 0 has a minimum value of 1 and a maximum value of 7 rank 1 has a minimum value of 14 and a maximum value of 23 Now what do you do when you get a value of 10? I think your entire concept is flawed, as empirical buckets don't have hard and fast upper and lower limits, they have empirical limits, as I have described, and will result in problems if you try to apply this to a 2nd data set. The idea of creating buckets from continuous variables is also a flawed idea, in my opinion, and you might want to re-think what you are doing with this data. If you explain the long-term analysis goal, as well as what the data is, we might be able to propose more meaningful analysis techniques.

PaigeMiller · ‎09-07-2012

Thanks, Tom, I knew it must be something simple.

PaigeMiller · ‎09-07-2012

Here's my existing code, that always gives me an error %let abc=%sysfunc(intnx('dtday',1662637746,365),datetime16.); %put ABC &abc; the large value 1662637746 is simply today's date and time in numeric format, which should work. I have tried other formats, including '07SEP12:00:00:00'dt, but I always get the same error message: WARNING: An argument to the function INTNX referenced by the %SYSFUNC or %QSYSFUNC macro function is out of range. NOTE: Mathematical operations could not be performed during %SYSFUNC function execution. The result of the operations have been set to a missing value. So what am I doing wrong?

PaigeMiller · ‎09-07-2012

A side comment ... if you knew the problem was reading the data, and you posted a topic/subject line that indicated you were having a problem reading certain type of data, you would get more of the right type of people reading this thread. As it is, your subject line attracts people who are statistical experts, but not experts at reading data problems. So, my suggestion is that you start a new thread with an appropriate subject/topic about how you are having trouble reading a certain type of data.

PaigeMiller · ‎09-07-2012

If you look at the SASLOG, you will find that your data set WORKERS never gets created because of errors in your code there.

PaigeMiller · ‎09-06-2012

You would use a MACRO variable here, see Example 1 in the %SYSFUNC statement documentation

PaigeMiller · ‎09-05-2012

PROC GLIMMIX or PROC GLM is definitely the way to go.

PaigeMiller · ‎08-31-2012

PTD_SAS wrote: I want to submit the statement ods path work.mystore(update) sashelp.tmplmst(read); when I invoke SAS so I can modify templates temporarily. I tried to include it in the configuration file sasv9.cfg but it doesn't work. Could I get some suggestions as to how I can achieve this? Very simple ... put it in your AUTOEXEC file. You may want to create a custom CONFIG file and AUTOEXEC file so the original remains unchanged.

PaigeMiller · ‎08-30-2012

Astounding wrote: Is there some reason you have to use PROC SUMMARY to accomplish this? PROC FREQ is built for this task: I would claim that PROC SUMMARY is also built for this task (as well as other tasks)

PaigeMiller · ‎08-30-2012

What does n = how_many represent ? How can I get "how_many" ? N=HOW_MANY creates a data set variable named HOW_MANY in SAS data set WANT that contains N (the count of the number of observations in each group). You access HOW_MANY the same way you access any data step variable.

PaigeMiller · ‎08-23-2012

This is not an appropriate use of Macro statements. You can accomplish everything you need in a data step, without ever using a macro. See the documentation for the LINK statement for examples on how to branch to specific code.

PaigeMiller · ‎07-27-2012

antony@allianz wrote: proc means data=y n; var id; run; proc sort data=y out=y1; by gender; run; data male female; set y1; if gender='m' then output male; else if gender='f' then output female; run; data x; set male; retain Mno; if first.gender='m' then Mno+1; if last.gender; run; data x1; set female; retain Fno; if first.gender='f' then Fno+1; if last.gender; run; data gp1 gp2 gp3; set x; Mcnt=Mno/31; g1=Mcnt*10; g2=Mcnt*11; do i=1 to Mno; if i<= g1 then output gp1; g4=g1+g1; else if g1 <i<= g4 then output gp2; g6=g4+g2; else if g4<i<= g6 then output gp3; end; run; data gp4 gp5 gp6; set x1; Fcnt=Fno/31; gf1=Fcnt1*10; gf2=fcnt*11; do i=1 to Fno; if i<= gf1 then output gp4; gf4=gf1+gf1; else if gf1 <i<= gf4 then output gp5; gf6=gf4+gf2; else if gf4<i<= gf6 then output gp6; end; run; then Append the datasets gp1 & gp4,gp2 & gp5 and gp3 & gp6 Obviously, there are many ways to solve this problem. I have a strong desire for the simplest solution whenever I write code, so this would fail on that score. The code posted by LinLin seems to be about as simple as it can get; and so I don't see any reason to use this code.

PaigeMiller · ‎07-27-2012

Fareeza Khurshed wrote: You can also use PROC RANK with a random variable, a touch easier than a dataset I believe. DATA strat_people ; INPUT participant gender $1. ; CARDS ; 1 M 2 F 3 M 4 F 5 M 6 F 7 M 8 F 9 M 10 F 11 M 12 F 13 M 14 F 15 M 16 F 17 M 18 F 19 M 20 F 21 M 22 F 23 M 24 F 25 M 26 F 27 M 28 F 29 M 30 F 31 M ; data strat_people; set strat_people; random=ranuni(88); run; proc sort data=strat_people; by gender random; run; proc rank data=strat_people out=results groups=3; by gender; var random; ranks randomgroup; run; You don't need two data steps to begin the program, this can be accomplished in a single data step.

PaigeMiller · ‎07-26-2012

Well, Linlin, nice job. Shame on me, I thought it would take two PROC SORTs ... duh ... but you use two datasteps after the SORT, I'm sure you could do it in one datastep

PaigeMiller · ‎07-26-2012

I could envision a much simpler program that doesn't require any macros at all, just a data step, two PROC SORT steps, and another data step.

Online Status	Offline
Date Last Visited	5 hours ago

Re: VALIDVARNAME=V7

Re: How to Reg on each row?! with Slope/Intercept saved out?!

Re: Programming a time-varying cumulative covariate in long format

Re: Programming a time-varying cumulative covariate in long format

Re: Programming a time-varying cumulative covariate in long format

Re: How to Reg on each row?! with Slope/Intercept saved out?!

Re: Programming a time-varying cumulative covariate in long format

Re: How to Reg on each row?! with Slope/Intercept saved out?!

Re: Programming a time-varying cumulative covariate in long format

Re: SGPLOT how to specify line color/thickness for a bunch of lines?!

Re: DateTime formatting

Re: DateTime formatting

Re: How to put the codes into a macro?!

Comparing Slopes from Separate Regressions

Re: wide or long structure

Re: How to Reg on each row?! with Slope/Intercept saved out?!

Re: problem with ODS in SAS EG 8.3

Re: problem with ODS in SAS EG 8.3

Re: INPUT not converting character to numeric

Re: SGPLOT How to set y2axis including zero?!

Re: Help - Using Proc Rank With Two Datasets

Re: Can't get INTNX to work inside %SYSFUNC

Can't get INTNX to work inside %SYSFUNC

Re: A regression problem...

Re: A regression problem...

Re: How do I put the current date into the name of a file to export?

Re: Reapeated measures linear regression

Re: Customising the SAS startup

Re: need help with summarizing data

Re: need help with summarizing data

Re: problem with using %goto in a data step

Re: how to divide dataset into 3 samples,each one with 50% males & fem...

Re: how to divide dataset into 3 samples,each one with 50% males & fem...

Re: how to divide dataset into 3 samples,each one with 50% males & fem...

Re: how to divide dataset into 3 samples,each one with 50% males & fem...

SAS Global Forum