/* sorry for my mistakes it works more easy with a data step but still understand why my macro program doesn't work , it seems the it is indefinitely runing .....*/   /* i have en error when i submit my code with my example where source and string contains quotation marks or special caracters   i have created the source and string macro variable with proc sql statment using into: separated by ' ' ( space ) */       %macro control_find_word(source,string,dlm);   　 %let k=1; %let ID_dim=%scan(&string. , &k ," "); %do %while (&ID_dim ne );   data _null_; result = indexw("&source","&string"," "); call symputx('result',result,'g');   run; %if &result=0 %then %do; %let control_dim=1; %put ERROR: WORD DOES NOT EXIST; %goto fin; %end; %let k=%eval(&k+1); %let ID_dim=%scan(&string. , &k ," "); %end;   %FIN : ; 　 %mend;   %control_find_word(&source,&string,);   ... View more

i am not talking avout SAS names but the values of the variables ... View more

oh ok . it is not possible to apply the same as in SQL meaning upcase in the 2 tables directly :(.   1. so i need to modify before the 2 tables for all characters variables concerned . i did'nt want to do this ,i was thinking that it could be possible in the HASH ( more efficient ? ) ...  i wanted to try something directly in the hash function.   2. there is more than 2 possibilities , hhh.find(key:upcase(BUSINESS_AREA)) only upcase in the second table ( table_ref2) , not in the set table .    if _n_=1 then do; declare hash hhh(dataset: "table_ref2 (keep=rate_1 BUSINESS_AREA )", multidata:'y'); hhh.DefineKey('BUSINESS_AREA'); hhh.DefineData(ALL:'YES');   hhh.DefineDone(); end; do until(DerniereObservation); set have2 end=DerniereObservation; call missing(rate_1);   rc = hhh.find(key:BUSINESS_AREA); rc2= hhh.find(key:lowcase(BUSINESS_AREA));     ... View more

Thanks for your responses,   i have many character variables in the key ( here an example with 2 )  and the case issue concern the 2 tables ( the orignal table and the ref table i want to join )  . for example i can have in the original table : All_buisness and in the ref table ALL_BUISNESS , so i need to apply either upcase/lowcase in the first but also in th second table   i found a solution by creating SAS datasets views ..... but do you know the impact in performance ... ( big datasets ...)   data table_ref2 / view=table_ref2; set table_ref;  BUSINESS_AREA = lowcase(BUSINESS_AREA); run;   data have2 / view=have2; set have;  BUSINESS_AREA = lowcase(BUSINESS_AREA); run;   data have3(drop=rc); /*iterate left data set*/ set have2 ; /*declare variables from hash set*/ length rate_1 rc 8 ; /*declare hash*/ if _n_=1 then do; declare hash hhh(dataset: "table_ref2(keep=rate_1 BUSINESS_AREA TYPE_OF_POL )", multidata:'y'); hhh.DefineKey('BUSINESS_AREA','TYPE_OF_POL'); hhh.DefineData(ALL:'YES');   hhh.DefineDone(); end; do until(DerniereObservation); set have2 end=DerniereObservation; call missing(rate_1);   rc = hhh.find(key:BUSINESS_AREA,key:TYPE_OF_POL); if rc ne 0 then do; var =.; output; end; do while (rc=0); var =rate_1*10; output;   rc = hhh.find_next(); end; end; run;   ... View more

Yes thanks it could be another solution , but the merge involve a sort of the 2 table , more time consuming i think ? ... View more

I have all controls before doing the join , i am sure i have the same data structure between tables ( same variables with same type and format ... ) . Do i need to generate by a loop the syntax : on a.first_item=b.item and a.second.item=B.item in a different macro variable then use it my sql code? ... View more

for my need ( create a file which is going to be call in a PROC HTTP ( in parameter)  ) , i used the JSON procedure ( i have never used it before ... ) :   http://documentation.sas.com/?docsetId=proc&docsetTarget=n0s97o94nhawspn1rjhhsiu4nucz.htm&docsetVersion=9.4&locale=fr   thanks for your help. ... View more

strat day can be any date the goal is to split by any anniversary dates the month and always have end of month. for example 15/month/year and 20/month/year i always want : start date / end date from the begining of each month until the 15 then 15 to 20 and finally 20 to the end of month . ... View more

thanks for you support , i have tested what i really want using you suggestion but i don't have the result expected ...   i have a second anniversary date corresponding to the birthday ( month slicing) ...   data input; input birth_date: ddmmyy10. start_date: ddmmyy10. end_date: ddmmyy10.; format birth_date start_date end_date ddmmyy10.; datalines;   06/07/2008 15/06/2008 31/12/2010 ; run; data expected; set input; format date_breaks ddmmyy10.; max_dur=intck('month', start_date ,end_date,'continuous'); do i=0 to max_dur; date_breaks=intnx('month', start_date,i,'s'); output; date_breaks=intnx('month', birth_date,i,'s'); output; end; drop i ; run; data want; set expected; format date_break1 date_break2 ddmmyy10.; if mod(_n_, 2) then do;   date_break1 = date_breaks; date_break2 = intnx('month', date_breaks, 0, 'end'); end; else do;   date_break2 = date_breaks; date_break1 = intnx('month', date_breaks, 0) - 1; end;   /*drop date_breaks;*/ run;   the result i have :   15/06/2008 30/06/2008 30/06/2008 06/07/2008 15/07/2008 31/07/2008 31/07/2008 06/08/2008 ..   but i want :   15/06/2008 30/06/2008 30/06/2008 06/07/2008 06/07/2008 15/07/2008 15/07/2008 31/07/2008 31/07/2008   06/08/2008 ...   could you explain me this part of the code :   if mod(_n_, 2) then do;   Many thanks ... View more