Are you trying to display numbers using decimals as the thousands separator?  If so, the very hacky way to do this is to just use the PUT() function to convert to a character string.  Normally, Excel will undo this (automatically interpreting anything that looks like a number as a number), but it won't do this in the case of a number formatted like this: 824.586.182,15 ...assuming you're opening it in a locale where the above is not the standard.     options locale=German_Germany; data test; num=198481788.01; numc=put(num, nlnum15.0); run; proc export data=test dbms=xlsx replace outfile="/home/xxxxx/testnum.xlsx"; run; ... View more

As an aside, you might want to consider use of "not quite a date" variable like your "AccountingYearMonth" with apparent plain numeric values of 202509. Use an actual date value that represents the first day of the month to allow use of the SAS functions that will deal with increments and intervals . You can assign a format to the values that would have a default appearance of YYYYMM. Your code accountingYearMonth LE 202509 would become  accountingYearMonth LE "01Sep2025"D which is, in my opinion, a bit clearer to follow the intent of the code.   SAS understands dates for many of the graphing procedures to do things like place tick marks at month or quarter intervals but graphing something with values like 202509 means a big gap between 202512 and 202601. Plus the modeling and analysis procedures can use the intervals. Not to mention simple things like the functions such as Quarter that would let you select a calendar quarter or a minor format change in a report could change the intervals from monthly to quarterly or annually ... View more

You can use Proc Format to tell you about the custom formats/informats you have available.  Example that creates one format and one informat in the Work library and default format catalog. Then  gets the descriptions with the FMTLIB option. Run the below code and see in the description the the $ABC has the words FORMAT NAME, indicating it is an output format and $PDQ has INFORMAT NAME indicating it is used for reading.   You would want to use Library=LIBRARY.cerform in the Proc format statement to examine you custom formats.    /* create a custom format to use later*/ proc format ; value $abc 'abc'='some formatted result' ; invalue $pdq 'pdq'='input result value' ; run; /* get descriptions of the current custom format/informats */ proc format library=work fmtlib; run;  I would quite often have an informat/format pair defined because the INVALUE has different options available such as <some value description> = _error_ to report in the log that the value encountered may not be valid and reports as such in the log, or to create special missing values that a FORMAT could then provide details about when requested. Plus for reading text the INVALUE option UPCASE is very helpful to standardize values so that Abc, ABc, aBc, abC (etc) are all read into the value ABC. ... View more

We can't know what your SAS scripts like HYFI_INTExpSF.sas are actually doing but if the create text files then it's highly likely that they use a SAS table as source to create these text files.  As others already mentioned Proc Compare works on SAS tables and not text files. If your current scripts create these text files as a 1:1 export from SAS tables AND if these SAS tables are permanent, then you could of course write another SAS script that uses Proc Compare and compares the source SAS tables.    ... View more

Yes you are right, that will do the same ... View more

Hello @alepage,   @alepage wrote: (...) if not missing(writtenUnit_HyFi_sas) and not missing(writtenUnit_harmony) then do; if writtenUnit_HyFi_sas eq 0 then wp_diff_pct = 0.0000; else wp_diff_pct = round((writtenPremium_harmony / writtenPremium_HyFi_sas) - 1,0.00001); end; run; It is good practice to check for missing values and zero denominators before dividing two numbers. Your code above has all the necessary IF-THEN/ELSE statements for those checks, but then surprisingly divides the values of two variables which were not checked in the IF conditions.   The code would look much more consistent -- and, most importantly, avoid the "division by zero" problem -- with either writtenUnit_harmony / writtenUnit_HyFi_sas in the division or writtenPremium_harmony and writtenPremium_HyFi_sas, respectively, in the IF conditions. ... View more

Try system options ERRORS=0 :   options errors=0; data x; set sashelp.class; x=input(sex,best.); run; 10 options errors=0; 11 data x; 12 set sashelp.class; 13 x=input(sex,best.); 14 run; NOTE: 达到 ERRORS= 选项设置的限制。系统将不打印该类型的其他错误。 NOTE: 在以下位置无法执行算术运算。运算结果已设为缺失值。 每个位置的指定方式: （次数）位置（行:列）。 19 次，位于 13:3 NOTE: 从数据集 SASHELP.CLASS. 读取了 19 个观测 NOTE: 数据集 WORK.X 有 19 个观测和 6 个变量。 NOTE: “DATA 语句”所用时间（总处理时间）: 实际时间 0.01 秒 CPU 时间 0.01 秒 ... View more

Not with the XPORT libref engine.   Did you try using the %XPT2LOC() autocall macro that SAS provides instead? filename myxport zip "/dwh_actuariat/sasdata_controlled/current/be/othr/bel.prod0030.saf3fm.data/saf3fm_billing.zip" member="SAF3FM.XPORT" ; %xpt2loc(filespec=myxport) If that doesn't work then try using this macro instead https://github.com/sasutils/macros/blob/master/xport2sas.sas %xport2sas(myxport) ... View more

Concur with what the others said. Set your editor so that it replaces the tabs with spaces whenever tabs are entered. This also prevents the effects of different tab-width settings on different systems. ... View more

Read for yourself: find  It's in the OPERATORS section. ... View more

I am a SQL guy, just for having some fun . data x; do year=2001 to year(date()); output; end; run; proc sql noprint; select year into :yearlist separated by ' ' from x; quit; %put &=yearlist.; ... View more

Two general solutions. Leave your data step essentially the same, but write the "APPEND" records to a temporary dataset.  Then use a second step running PROC APPEND to add that temporary dataset to the permanent dataset.   Write the matching records to a dataset, let's call it BOTH.  Write the unmatched BASE records to another temporary dataset, let's call it TEMP_BASE.  Then append that temporary dataset to some existing (or not existing dataset since PROC APPEND does not care) permanent dataset, let's call that PERMANENT_BASE.   data BOTH TEMP_BASE ; merge BASE (in=inbase) LIST (in=inlist where=(filename = "MYFILENAME")) ; by BYVAR ; if inlist and inbase then output BOTH; else if inbase then output TEMP_BASE; else /* Ignore the LIST_ONLY records */ ; run; proc append data=TEMP_BASE base=PERMANENT_BASE; run; NOTE: Once you know what code you want to generate you can start replacing some of the code with macro variable references.   The other way is to use a MODIFY statement which will allow you to use the APPEND statement to append observations to a dataset in a single data step.  You can also use an OUTPUT statement to write those other observations to some other output dataset.   It is hard to tell from your dataset names and your descriptions so far exactly what you are trying to do.  But perhaps it is something like this:    You have two input datasets named BASE and LIST and you want to combine them to generate observations that are either appended to PERMNENT or written to a brand NEW dataset.     data permanent new ; if 0 then modify permanent; merge base(in=inbase) list(in=inlist); by keyvar; if inbase and inlist then output new; else if inbase then append; run;   But somehow I doubt it.  More likely you will want to use the KEY= option on the MODIFY statement to get it to work.   Providing a simple example of the inputs with just enough observations to demonstrate the different possible situation will make it much easier to design and test different solutions.     ... View more

I have added end=eof and the if eof and _infile_=' ' then delete; at the end of the input statment as you did and it works fine ... View more