One way you could try is: 1) Run PROC CPORT to export your IML library into a XPT file. 2)Download this XPT file. 3)Use PROC CIMPORT to import this XPT file at your Windows PC side. ... View more

The following code is changing all the .TXT files under a fold into .SAS files. %let path_in= D:\temp\02_table ; *the pathname for txt files; %let path_out= D:\temp\02_table ; *the pathname for sas files; filename x pipe "dir &path_in.\*.txt /s /b"; data x; infile x truncover; input old $2000.; new=cats(prxchange('s/\w+$//',1,strip(old)),'sas'); rc=rename(strip(old),strip(new),'file'); run; And an alternative code by using sas capability: (especially your file name is utf8 encoding) /******For UTF-8 编码的文件名***********/ %let path_in= D:\temp\02_table ; *the pathname for txt files; %let path_out= D:\temp\02_table ; *the pathname for sas files; data _null_; length fname old new $ 200; rc=filename('x',"&path_in."); did=dopen('x'); do i=1 to dnum(did); fname=dread(did,i); if lowcase(strip(scan(fname,-1,'.')))='txt' then do; old=cats("&path_in",'\',fname); new=cats(prxchange('s/\w+$//',1,strip(old)),'sas'); rc=rename(strip(old),strip(new),'file'); end; end; run; ... View more

Yes. You are unable to include CARDS or DATALINES statement in a macro. This limited thing has been mentioned in sas documentation.  You would either include these data in a file or hard code these data like:   %macro init_er(); data _xtemp; endrange=-100 ; output; endrange=-75 ; output; endrange=-50 ; output; ; run; %mend; %init_er(); ... View more

@Rick_SAS  That is what I suggested. I meet such kind of problem before. The reason why I suggest this code is that I  notice there is a WARNING msg under ERROR, that WARNING would appeared if the data is not good enough to fit this model.   Maybe this data is not suited for PROC REG. For example, this dataset has only one or two obs ,that is not enough to fit a REG model.  %if &nobs > 10 %then %do; That might suppress these ERROR/WARNING message. Just could give it a try  I think.     ... View more

I noticed that you have WHERE condition : proc reg data=&ds.(where=(ind between &stind.+&stoff. and &endind.-&endrange.)); Maybe That is where reason from . When You add IF or WHERE condition ,that would lead to an empty dataset ,and that empty dataset would generate this WARNING , I think. So I suggest you to write some code to check this dataset is empty or have enough obs to do PROC REG. Like: data _ds; set &ds.(where=(ind between &stind.+&stoff. and &endind.-&endrange.)); run; %let dsid=%sysfunc(open(_ds)); %let nobs=%sysfunc(attrn(&dsid,nlobs)); %let dsid=%sysfunc(close(&dsid)); %put &=nobs. ; %if &nobs ne 0 %then %do; ods exclude all; ods output ParameterEstimates = _index_avg_parms; proc reg data=_ds; by tick; model _ret_t=ind; run; %end; ... View more

If you want to compare the mean from A table and B table, then it has nothing to do with Paired-T test. You could just use simple T test or non-parameter version T test (Wilcoxon Test - Check PROC NPAR1WAY) . But for you this special case, you have a YEAR variable, that seems it is a panel data or repeated measure data. You need to use Mixed model, check PROC MIXED or PROC GLIMMIX . And if you have SAS/ETS module, also could use PROC PANEL , but that is a Time Series Analysis issue, and post your question at Forecasting Forum: https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/bd-p/forecasting_econometrics ... View more

Then OP also need to control the max value(check its max >0 or <0) ? But OP only mentioned the ranges [-5,20] and [5,20]. ... View more

No need it I think. If the min value less than 0 , &min. would be its min value. Otherwise, &min would be 0. Of course, you could make it more complicated if you need to control MAX value too. ... View more

Yeah. Agree. It is all depend on what OP want. This example maybe is just an example, only OP know the real demand and real data . ... View more

As Tom pointed out ,using strip to get rid of trailing blanks: pos1 = find(string, strip( char_substring )); Otherwise ,sas would take 3000 as 3000blankblank..... Not 3000 ... View more

If these files have the same data constructure, that would be easy by FILEVAR= option.   data path; input path $80.; cards; c:\temp\a\PRJ__20221201_13227_MF.txt c:\temp\a\PRJ__20221201_13228_MF.txt ; data MF; set path; length from_file file $ 80; infile dummy filevar=path end=last filename=file truncover; from_file=file; do until(last); input a $ b $ c $; output; end; run;       If you don't like to type file name by hand, try this one:   /* data path; input path $80.; cards; c:\temp\a\PRJ__20221201_13227_MF.txt c:\temp\a\PRJ__20221201_13228_MF.txt ; */ %let path= c:\temp\a ; data path; length path $ 200; rc=filename('x',"&path."); did=dopen('x'); do i=1 to dnum(did); file=dread(did,i); if prxmatch('/.+MF\.txt$/i',strip(file)) then do; path=catx('\',"&path.",file); output; end; end; keep path; run; data MF; set path; length from_file file $ 200; infile dummy filevar=path end=last filename=file truncover; from_file=file; do until(last); input a $ b $ c $; output; end; run; ... View more

Also could check PROC MIXED or PROC GLIMMIX . ... View more

Yes. Make a macro to deal with it.   proc sql noprint; select ifn(min(ind_off25_min)>0,0,min(ind_off25_min)) into :min trimmed from _dif_temp_coll; quit; %put &=min; proc sgplot data=_dif_temp_coll; series x=endrange_n y=ind_min_dif / lineattrs=( color=blue thickness=2 pattern=solid); series x=endrange_n y=ind_off25_min / y2axis lineattrs=( color=red thickness=2 pattern=solid); refline &endind./axis=y lineattrs=(color=lime thickness=1 pattern=solid); refline 0/axis=x lineattrs=(color=lime thickness=1 pattern=solid); y2axis min=&min. ; run;quit; ... View more