About MikeZdeb

MikeZdeb · ‎08-30-2011

Hi ... you got an answer from Art, but given that you tried ... name2=reverse(substr(left(reverse(name)),5)); here's another function idea. Just make the format in the PUT statement 4 less than the length of NAME data x; input name $20.; datalines; SAW_PRICE_STD XYZ_STD MIKE_STD ABCDEFGHI_STD ; run; data x; set x; name = left(put(right(name),$16.)); run; proc print data=x; run; name SAW_PRICE XYZ MIKE ABCDEFGHI

MikeZdeb · ‎08-29-2011

Hi ... here's another idea ... DATA CDR; INPUT phone1 phone2 date:date9. phone1_status:$1. phone2_status:$1.; format date date9.; CARDS; 1 2 01JAN2007 A B 1 3 02JAN2007 A B 2 3 03JAN2007 A B 2 4 04JAN2007 C A 4 1 05JAN2007 C B RUN; proc sql; create table new as select * from (select phone1 as phone, phone1_status as status, date from cdr union select phone2 as phone, phone2_status as status, date from cdr) group phone having date eq max(date); quit; proc print data=new noobs; run; phone status date 1 B 05JAN2007 2 C 04JAN2007 3 B 03JAN2007 4 C 05JAN2007

MikeZdeb · ‎08-27-2011

Hi ... you have suggestions on how to correct the error. Why did you get this message ... The range is repeated, or values overlap You only put part of the message in the posting ... ERROR: This range is repeated, or values overlap: .-.. With the format name GENDER, you asked to create a numeric format. PROC FORMAT sees start value is 'M' (character) and tries to do a character-to-numeric conversion ... that makes the start value missing. When there's no end value specified, end is set to the same value as start and the range is . - . . The same thing happens when start is 'F' ... the range is again .-. ... a repeated range. If you experiment a bit ... data gender; retain fmtname 'gender'; input start $ label $; datalines; M MALE ; run; proc format library=work cntlin=gender; select gender; run; actually works and creates a numeric format ... ---------------------------------------------------------------------------- | FORMAT NAME: GENDER LENGTH: 4 NUMBER OF VALUES: 1 | | MIN LENGTH: 1 MAX LENGTH: 40 DEFAULT LENGTH: 4 FUZZ: STD | |--------------------------------------------------------------------------| |START |END |LABEL (VER. V7|V8 27AUG2011:00:18:47)| |----------------+----------------+----------------------------------------| | .| .|MALE | ---------------------------------------------------------------------------- as does ... data gender; retain fmtname 'gender'; input start $ label $; datalines; 1 MALE 2 FEMALE ; run; proc format library=work cntlin=gender; select gender; run; ---------------------------------------------------------------------------- | FORMAT NAME: GENDER LENGTH: 6 NUMBER OF VALUES: 2 | | MIN LENGTH: 1 MAX LENGTH: 40 DEFAULT LENGTH: 6 FUZZ: STD | |--------------------------------------------------------------------------| |START |END |LABEL (VER. V7|V8 27AUG2011:00:20:04)| |----------------+----------------+----------------------------------------| | 1| 1|MALE | | 2| 2|FEMALE | ---------------------------------------------------------------------------- In this case, the character-to-numeric conversion results is two, non-repeating ranges.

MikeZdeb · ‎08-24-2011

Hi ... for future reference, if ever in the situation where you have a data set, you can read the last observation directly ... data test; input seq @@; datalines; 1 2 3 4 5 6 7 8 9 999 ; run; data _null_; set test nobs=obs point=obs; put "last seq: " seq; stop; run; last seq: 999

MikeZdeb · ‎08-24-2011

Hi ... original posting mentioned 7 variables, plus as Art said ... the limit is the max length of a character variable, 32K+.

MikeZdeb · ‎08-23-2011

Hi ... here's another idea. Use PROC SUMMARY to aggregate the values within a vakue of KEY, then use a data step to concatenate the values into one variable. The CLASS variable avoids the sort. data input; input key $ data $; cards; Abc123. 077123 Abc123. 098546- Abc123. 875433 Def123. 345423 Def123. 344565x ; run; proc summary nway data=input missing; class key; output out=new (drop=_type_ _freq_) idgroup(out[3](data)=); run; data new; set new; newvar = catx(' ', of data:); keep key newvar; run; If you are not sure of the number of observations within each key, you can use ... proc sql noprint; select max(obs) into :obs from (select count(*) as obs from input group by key); quit; prior to PROC SUMMARY and substitute [&obs] for the [3] used above; fyi ... "Transposing Data Using PROC SUMMARY'S IDGROUP Option" http://www.lexjansen.com/pharmasug/2009/tt/tt08.pdf

MikeZdeb · ‎08-23-2011

Hi ... here's another idea, works no matter how many answers you have ... data x; input (answer1-answer4) (: $3.); datalines; Yes Yes No Yes No No Yes Yes No No No No Yes Yes No No Yes Yes Yes Yes No Yes No No ; run; data x; set x; new = ifc(countc(catt(of answer:),'Y') ge 2, 'YES' , 'NO'); run; proc print data=x noobs; run; answer1 answer2 answer3 answer4 new Yes Yes No Yes YES No No Yes Yes YES No No No No NO Yes Yes No No YES Yes Yes Yes Yes YES No Yes No No NO fyi ... "Searching for Variable Values with CAT Functions: An Alternative to Arrays and Loops" http://www.nesug.org/Proceedings/nesug09/ff/ff04.pdf

MikeZdeb · ‎08-23-2011

Hi ... I guess it's "log before you leap".

MikeZdeb · ‎08-23-2011

Hi ... closer to the data set in the original posting ... took about 25 seconds CPU (elapsed about the same) on a not too fast PC. * fake data ... 100,000 obs / 600 numeric + 200 character vars data x; array nn(600); array cc(200) $2.; do id=1 to 1e5; do k=1 to 200; nn(k) = ifn(ranuni(999) le .05 , . , 10); cc(k) = ifc(ranuni(999) le .05 , ' ' , '10'); end; do k=201 to 600; nn(k) = ifn(ranuni(999) le .05 , . , 10); end; output; end; keep id nn: cc: ; run; options missing=''; proc transpose data=x out=new (where=(missing(col1)) keep=_name_ col1); by id; var _all_; run; options missing='.'; proc freq data=new order=freq; table _name_ / nocum nopercent; run; some results ... _NAME_ Frequency nn314 5266 cc198 5230 nn124 5213 nn511 5185 cc182 5178 nn392 5169 nn3 5159

MikeZdeb · ‎08-23-2011

Hey Art ... pretty sure PROC MEANS only accepts numeric variables in the VAR statement. I get this with your example ... 183 proc means data=have noprint; 184 var _all_; ERROR: Variable var2 in list does not match type prescribed for this list. 185 output out=temp (drop= _type_ _freq_) nmiss=; 186 run; How about (assumes there is a variable that uniquely identifies each observation ... here it's NREC) ... data x; input name : $10. age zip : $5. salary; nrec+1; datalines; mike 64 12203 9999 sara . 20001 . . 24 20001 10000 jessica . 90124 999 ; run; options missing=''; proc transpose data=x out=new (where=(missing(col1)) keep=_name_ col1); by nrec; var _all_; run; options missing='.'; proc freq data=new; table _name_ / nocum nopercent; run; _NAME_ Frequency age 2 name 1 salary 1

MikeZdeb · ‎08-12-2011

Hi ... not an FTP answer, but just curious as to why you create the intermediate data set FTPDATA. Can't you just read/write within the same data step. I assume that the variable name FULLREC means read/write the entire record ... filename create ftp "&ftp_file_name" cd=&path host=&host user="&userid" pass="&pwd" recfm=v debug lrecl=&reclth; data _null_ infile inputfl; input; file create; put _infile_; run;

MikeZdeb · ‎08-10-2011

Hi ... here's another approach that works with your multiple moms. It uses a format as a lookup table to assign the PERSONIDs as a mother or father ... data one; input dwellingid : $4. personid : $3. id : $8. (momid dadid) (: $3.); datalines; 1001 101 1001_101 N/A N/A 1001 102 1001_102 N/A N/A 1001 103 1001_103 N/A N/A 1001 104 1001_104 101 102 1001 105 1001_105 103 102 ; run; proc sql; create table momdad (where=(start ne 'N/A')) as select distinct momid as start, '10' as label, '$momdad' as fmtname from one union select distinct dadid as start, '01' as label, '$momdad' as fmtname from one; quit; proc format cntlin=momdad; run; data two; set one; mother = ifc(momid eq 'N/A', char(put(personid,$momdad.),1), '0'); father = ifc(dadid eq 'N/A', char(put(personid,$momdad.),2), '0'); run; proc print data=two noobs; run; dwellingid personid id momid dadid mother father 1001 101 1001_101 N/A N/A 1 0 1001 102 1001_102 N/A N/A 0 1 1001 103 1001_103 N/A N/A 1 0 1001 104 1001_104 101 102 0 0 1001 105 1001_105 103 102 0 0

MikeZdeb · ‎08-10-2011

Hi ... there's always "y" and "z" once "x" reaches a certain size ... or your hash object.

MikeZdeb · ‎08-09-2011

Hi ... another idea with just the CATX and FINDW functions ... data temp; input city $8. ; cards; hyb vja nlg mum pune pune hyb vja triven run; data x; length x $1000; retain x; set temp; select; when (missing(city)) do; x = catx(' ',x,'miss'); nc+1; count=nc; end; when (^findw(x,trim(city))) do; x = catx(' ',x,city); nc+1; count=nc; end; otherwise count = findw(x,trim(city),' ','e'); end; keep city count; run; proc print data=x noobs; run; city count hyb 1 vja 2 nlg 3 4 mum 5 pune 6 pune 6 7 hyb 1 8 vja 2 9 triven 10

MikeZdeb · ‎08-08-2011

Hi ... sorry, no. I think that both a macro and call execute add a level of complexity to all this that you avoid using a data step to write SAS code to a temporary file that is included later.

Online Status	Offline
Date Last Visited	‎06-28-2018 10:46 AM

Re: Create variables as per separators in string

Re: How to create sub-totals ?

Re: Running a function for each unique value

Re: Aggregating dataset by date, including missing dates

Re: Aggregating dataset by date, including missing dates

Re: Aggregating dataset by date, including missing dates

Re: Aggregating dataset by date, including missing dates

Re: How to write generic libname to multiple databases?

Re: delete/retein observations by frequency (n) of variables

Re: delete/retein observations by frequency (n) of variables

Re: is there simpler way to convert char flag to numeric number

Re: INPUT function requires a character argument

Re: Renaming all variables in a table

Re: Report on missing values of each coloumn

Re: Remove the last character of my string

Re: Proc GMap: Caption and custom legend

Re: Saving PROC GPLOT graph to folder

Re: re: Catx Function Error

How to calculate geodesic distance in SAS

Re: Removing a String from a Char Variable

Re: In search of an efficient way to deal with ID<>ID records

Re: This range is repeated, or values overlap

Re: Reading the last record of a large table

Re: variables combination

Re: Combine multiple observations into one if they have the same Id

Re: variables combination

Re: Counting Missing Values

Re: Counting Missing Values

Re: Counting Missing Values

Re: Unable to use FILENAME FTP in production ZOS region.

Re: Creating a variable based on subsequent observations

Re: Reg Generating Seq no

Re: Reg Generating Seq no

Re: Proc Corr in a data step