Category Specific Quartiles

[kaitlynanderson]

I have a homework and I am being tasked with creating "sex specific quartiles". Here are the exact instructions

Create sex-specific quartiles for the variables “zdelta_muac” and “zdelta_bmi” (using the “visitdata4” dataset) and call the new ranked variables “q4s_zdelta_muac” and “q4s_zdelta_bmi.” Output these results into a new temporary dataset called “visitdata5.”

where zdelta_muac and zdelta_bmi are standardized variables. 

/*sex-specific quartiles*/
proc sort data=visitdata4;
by sex;
run;
proc rank data=visitdata4 out=visitdata5;
by sex;
var zdelta_muac zdelta_bmi;
ranks q4s_zdelta_muac q4s_zdelta_bmi;
run;

This is the code I have, but when I run this, it deletes a bunch of observations from visitdata4.  How do I make this not happen

[ballardw]

To do quartiles I would expect to see the option GROUPS=4 on the Proc Rank.

Do you have missing values for your BY variables?

Normally Proc Rank doesn't remove observations. Can you show us the LOG where you use Proc Rank to create Visitdata5?

Copy the text from the Log with the code and all the notes or messages. On the forum open a text box using the </> icon above the message window and paste all the text.

That would look something like this (your variables and sets differ of course)

3694  proc rank data=work.class groups=4 out=work.classrank;
3695     by sex;
3696     var height weight;
3697     ranks h_rank w_rank;
3698  run;

NOTE: There were 19 observations read from the data set WORK.CLASS.
NOTE: The data set WORK.CLASSRANK has 19 observations and 7 variables.
NOTE: PROCEDURE RANK used (Total process time):
      real time           0.01 seconds
      cpu time            0.01 seconds

[kaitlynanderson]

I forgot the groups = 4 and added that in. Here is a snippet of what my visitdata4 looks like. 

Then I run this code: 

/*sex-specific quartiles*/
proc sort data=visitdata4;
by sex;
run;
proc rank data=visitdata4 out=visitdata5 groups =4 ;
by sex;
var zdelta_muac zdelta_bmi;
ranks q4s_zdelta_muac q4s_zdelta_bmi;
run; 

and it now visitdata 4 looks like this:

 

Here is the log for the proc rank:

NOTE: There were 197 observations read from the data set WORK.VISITDATA4.

NOTE: The data set WORK.VISITDATA5 has 197 observations and 25 variables.

NOTE: PROCEDURE RANK used (Total process time):

real time 0.00 seconds

user cpu time 0.00 seconds

system cpu time 0.00 seconds

memory 2831.28k

OS Memory 23724.00k

Timestamp 04/23/2024 06:59:36 PM

Step Count 99 Switch Count 4

Page Faults 0

Page Reclaims 251

Page Swaps 0

Voluntary Context Switches 20

Involuntary Context Switches 0

Block Input Operations 0

Block Output Operations 528

thanks for the help!

[ballardw]

Your log bit

NOTE: There were 197 observations read from the data set WORK.VISITDATA4.
NOTE: The data set WORK.VISITDATA5 has 197 observations and 25 variables.

Confirms than no observations were deleted.

If you are showing Visitdata4 (visitdata5 would look the same), it is showing the SORT order from your sex variable. Sex is missing for 5 observations. The places where sex is missing seems to be associated with missing other variables as well.

I bet a small pile of $$$ that if you run something like this (or with any of your Visitdata1, Visitdata2 guessing these exist) then you will see that Sex is missing for 5 observations.

proc freq data=visitdata3;
   tables sex / missing;
run;

The ranks for observations where the By variable is missing will be missing as well.

[kaitlynanderson]

This makes sense now! Thank you for the explanation I appreciate it!!