@Kurt_Bremser wrote: @SASPhile wrote: there are 100's of variables in multiple datasets Which means you have a serious conceptual issue. You should take care of that when you import the data in the first place.   Quite so. I have several custom informats depending on the project that turn various text strings such as N/A, UNK, or  "(NO DATA)" and "(NOTHING ENTERED)" to missing, or "special missing" as needed, when reading the files. ... View more

@deepakkailay this thread is 4 years old. If you have a new inquiry: Search support.sas.com and this site so that your question have not already been answered If no hit, create a new subject/thread ... View more

The basic approach I would take would be along these lines: data want; set have (rename=(bu=buold)); do i=1 to countw(buold); bu = scan(buold,i); output; end; drop i buold; run; It would likely be a good idea to set a length for the BU variable to longest expected value but I don't know what that might be so leave it up to you.   ... View more

You mean this ?   data have; x='input(Oldvar,best.)'; want=tranwrd(x,',','_,'); run; ... View more

Works fine for me. %let i=1 ; %let var1=bank; %macro attrib_bank; * part of a data step ; %mend attrib_bank; options mprint; data test_bank; %attrib_&&var&i; run; 17 data test_bank; 18 %attrib_&&var&i; MPRINT(ATTRIB_BANK): * part of a data step ; 19 run; NOTE: The data set WORK.TEST_BANK has 1 observations and 0 variables. ... View more

@SASPhile wrote: I'm not looking if a variable exists. I'm looking for a variable that contains TEST as part of variablename and if it has at least one missing values. HOW many variables are you concerned with in each data set? What does the output potentially look like with 10 variables? And are you really sure that you want every variable that may have 'TEST' as part of the name? What if you have a variable named "testimony" "intestate" Do you want each time a variable is missing or if any records in the set are missing values at least once?   ... View more

@ybolduc You need to be careful which string function you're using as soon as it comes to dealing with multi byte characters.  The PRX...() functions are unfortunately only good for single byte. http://support.sas.com/documentation/cdl//en/nlsref/69741/HTML/default/viewer.htm#p1pca7vwjjwucin178l8qddjn0gi.htm  ... View more

And if anyone earlier made a comment about "why stick multiple values into a single value" you just discovered one of the reasons the question was asked. ... View more

I used compress(dr3,,'kw') and it worked. (keep only printable characters)! Thanks again for your help! ... View more

A slightly fleshed out version of @tomrvincent's suggestion:   data sortedmeds; set meds; call sortc (dgr1,drg2,drg3); run; proc sql; create table want as select a.*, b.name from sortedmeds as a left join regiment as b on a.drg1=b.drg1 and a.drg2=b.drg2 and a.drg3=b.drg3 order by a.id; quit; If the REGIMEN data is not sorted by name for the drg variables follow the pattern in the sorted meds.   I created a different data set just in case some other processing of meds data requires the original order. ... View more

@SASPhile wrote: two columns colA and Colb are present in dataset have and both are numeric. When I use the following I'm getting Error "A Character operand was found in %eval function.   %let var1=%eval((colA+colB)*colB));   data test;   set have; va1=&var1.; run; Once again, catastrophic misunderstanding of macro timing. Lots of insights into the force of macro to gain you need, young padawan, before use it you can without harm.   The macro PREprocessor sees only text. The text (colA+colB)*colB is nothing numeric that can be evaluated, so the %eval (as it happens immediately in the %let) will fail. I also fail to see the wisdom of packing data step logic into a macro variable. This will only reduce the maintainability of your code. ... View more

You can find the names and properties of variables in Dictionary.columns if using Proc Sql (probably best for this exercise) or sashelp.vcolumn.   For instance the following code gets the variables from SASHELP.CLASS. proc sql; select * from dictionary.columns where libname='SASHELP' and memname='CLASS' ; quit; Caveats: the libname and memname (dataset name) are stored in uppercase but the name (variable name) may be mixed case so you need to address that. If you have as many libraries and data sets as I do you might want to create a subset dictionary.columns to a single library if you are going to use it repeatedly.     And here is brief example of comparing things: data work.class; set sashelp.class; drop age; run; proc sql; select * from (select upcase(name) as name from dictionary.columns where libname='SASHELP' and memname='CLASS') except select * from (select upcase(name) as name from dictionary.columns where libname='WORK' and memname='CLASS') ; quit; Your data set would likely take the place of the bit above where I extract names from dictionary.columns   Since you have a data set name then you would also need to match on the name instead of the explicit memname I used. You would need to provide mapping from "business_unit" to terms of SAS libraries or data sets. ... View more

if the number is 0.25, the decimal is not captured. shows as 25 if there numbers are like .63-0.125 mg,  .63-0.125 are not captured.,shows as 125 if there is no space between digit and unit like 0.25Mg nothing is captured ... View more