I was able to get to my desired results by adding another step proc sql; create table final as select *, count(id) as cnt from want group by id having calculated cnt >=5 ; quit;   Does anyone know if it is possible to do it in one step? ... View more

Hello FreelanceReinhard,   I have a similar issue but the code provided is not solving it. I want the 2 records before and 2 records bracketing a non missing FIRSTE value. So if there aren't  two records before or two records after the non missing value of FIRSTE then all the records for that ID get deleted. In the data below only the records from C and D should be in the final dataset but with the code   data have; input ID $ AGE RATE TYPE $ FIRSTE; datalines ; A 14.1 -4.19560471 N . A 15.1 -4.186550471 E -4.186550471 A 15.3 -4.041576561 E . A 15.6 -3.984641999 E . A 15.8 -4.079763059 E . A 16.2 -4.198425079 E . B 6.2 -1.118603619 N . B 12.25 -3.251832999 N . B 12.4 -3.158418124 N . B 14.8 -3.739770279 N . B 19.1 -3.980687937 E -3.980687937 B 19.6 -3.901730725 E . C 4.3 2.1339735869 N . C 5.63 2.1571201461 N . C 5.6 2.1571201461 E 2.1571201461 C 6.1 2.6609995525 E . C 6.7 2.1939735869 E . C 7.1 2.6991891929 E . D 2.4 -0.35998761 N . D 2.6 -0.650544267 N . D 3.2 -0.817670577 E -0.817670577 D 3.7 -0.659657975 E . D 4.2 -0.370159291 E . ; run; data want; set have; if firste>. then do p=max(_n_-2,1,p) to min(_n_+2,nobs); set have point=p nobs=nobs; output; end; run;     How can I eliminate records which do not match my condition? thanks ... View more

I would like to do a proc means  on the variable age and get the median, qand the 25 and 75 percentile where each of the flags are equal to 1. I can do it with a where condition in 3 steps    the flags are numeric, I made a typo previously input id $ age hip  knee  both; datalines; AB01 55 1 0 0 AB01 65 0 1 1 AC01 65 0 1 0 AC01 70 0 1 0 AC01 75 1 0 1 AD01 50 0 1 0 AD01 60 0 1 0 ;   proc means data=have noprint; where hip=1; var age; output out=want1(drop=_type_ _freq_) median= p25= p75= / autoname ; run;   proc means data=have noprint; where knee=1; var age; output out=want2(drop=_type_ _freq_) median= p25= p75= / autoname ; run;   proc means data=have noprint; where both=1; var age; output out=want3(drop=_type_ _freq_) median= p25= p75= / autoname ; run;   Can I do it in one step , where  I get the median and the percentiles of the age variable when each flag =1? I would get one observation: median_hip p25_hip p75_hip      median_knee p25_knee p75_knee      median_both p25_both p75_both ... View more

Thank you everyone. Too bad this functionality doesn't exist. I will move the text file to excel, update them in excel and then do a SAS import.  ... View more

Hi Paige, I am using the enhanced editor. I am not looking to create an abbreviation for propcase. I want to create a keybord shortcut which transforms the select text to propcase. If I copy my la my labels from a word.doc file into SAS instead of manually updating them I would like to be able to select the text and press CTL+SHT+alphaKey to have the text converted to propcase just like CTL+SHT+L does when it allows the user to convert the selected text in lower case. ... View more

Hi Diana,   if you add an id  t your dataset you can do a proc transpose to have all the values in one column.   data have; input fav_color1 fav_color2 fav_color3 fav_color4; datalines; 2 8 . . 1 3 7 9 4 9 . . 1 8 . . 2 . . . run;   data want ; set have ; id=strip(put(_n_,8.)); run;   proc transpose data=want out=tr_want(rename=( col1=color)); by id; var fav_color1 fav_color2 fav_color3 fav_color4 ; run;   proc freq data= tr_want noprint; tables color/missing out=test; title "Frequencies of color "; run; ... View more

The retain would not work because we want the closest measurement within a 3 week window so for this patient; 001 65 160 09MAR2017 001 65 . 06JUN2017 001 65 160 18JUL2017 for the record where height is missing (06JUN2017) we would want to pick the measurement on 18JUL2017 and not the one from 09MAR2017 because the 18JUL2017 is closer to 06JUNE2017 than 09MAR2017 ... View more

Thank you it worked ... View more

No this is not correct because ID2 has some ID1 which are in the column ID1. I only want the total of IDs which are unique to ID1 ad ID2, which would be (A,B,C,D,E,F,G,J ,I ,K,L,M,P,O= count=14).   The sum will sum the unique ID1 and Unique ID2 ... View more