Hi, In this case, you need to fix how many instances of step_cd and col_group...If there's so many subjects of step_cd and col_group then only you should think about Macros or formats... Based on SAS Codes, it will be difficult to answer your question so please post some sample data and output you want... -Urvish ... View more

Hi, Here is my try...Might be it will same as others' logic... data have(drop = position length);   input test_var $50.;   length state $5. city $20.; state = scan(test_var,-1," ");   call scan(test_var,-1,position,length); city = substr(test_var,1,(position-1)); cards4; EAST TEMPLETO      MA ST. PETE           FL ;;;; -Urvish ... View more

Hi, ODS OUTPUT Association = Association_Stat; Your PROC LOGISTIC STEP... ............... ............... ODS OUTPUT CLOSE; Make Sure that your Proc Logistic doesn't include NOPRINT option...Otherwise it won't create the dataset... -Urvish ... View more

Hi, Here is my try...Hope it meets the requirement... /*In Two Step*/ proc sql;    create table reg_data(drop = temp) as    select a.*,year(a.data) as temp,count(calculated temp) as tot_days    from pop as a,pop as b    where a.data= b.data    group by a.year    having tot_days GE 200    order by a.data; quit; proc sql;    create table reg_data as    select a.*    from reg_data as a right join sample as b    on a.firmid = b.firmid; quit; /*In one step*/ proc sql;    create table reg_data(drop = temp) as    select c.*,year(a.data) as temp,count(calculated temp) as tot_days    from pop as a        left join         pop as b    on a.data = b.data right join         sample as c    on a.firmid = c.firmid    group by a.year    having tot_days GE 200    order by a.data; quit; -Urvish ... View more

Hi, Looking forward, see the following data structure where your master dataset contains consecutive duplicates as well as non matching values compare to your transaction dataset... Neither the techniques posted above can achive the desired output of applying each update for each duplicates in master dataset...Not even UPDATE statement of PROC SQL posted in your first question will work here...So dont stuck on only one SAS techniques... See the following modified code... /*===================================================*/ /* Master Data contains consecutive duplicate values */ /*===================================================*/ data master;   input name $ value;     cards4; A 10 A 20 B 30 D 40 A 30 D 20 ;;;; /*============================================================*/ /* Transaction Data contains the values to be updated         */ /* with some new values which are not there in master dataset */ /*============================================================*/ data trans;    input name $ value;    cards4; A 80 C 30 D 50 ;;;; /*==============================================================*/ /* Data Step won't work to apply the update for each duplicates */ /* in master dataset, niether the COALESCE will work for this   */ /* kind of situation, all we need to use CASE expression to     */ /* apply correspondence update in master dataset                */ /*==============================================================*/ proc sql;     create table want(drop = value rename = (value1 = value)) as     select coalesce(a.name,b.name) as name,          case              when a.value NE b.value then b.value                             else a.value                    end as value,          case              when calculated value = .then a.value                             else calculated value         end as value1   from master as a             full join             trans as b   on a.name = b.name   orderby 1; quit; -Urvish ... View more

Well there're so many ways of playing with in SAS Datastep...so it will be good if you post how your transaction and master dataset looks like... The code i have posted, it works if you want to have update on single variable... In case of more variables, you only need to add them in RENAME = dataset option and let the update will occure for all the observations... As you said datastep update is capable of one to one update only...it is very true as per the SAS Documentation...However, it is also mentioned that if your master dataset is having duplicate values you can use this kind of tecniques...There is always limitation of any SAS Statements,options,functions,etc...So i think we should use all together, rather than stuck on only UPDATE statement, in such a way that we will have our desired output... The all you need is like you only need to rename the variables if there is more than one variables...See the following code... proc sql;   create index name   on new_data(name); quit; data want(drop = ori_val ori_val1);   do until(totobs);        set main_data(rename = (my_value = ori_val my_value1 = ori_val1)) nobs = totobs;        set new_data key = name;        if _iorc_ NE 0 then do;        my_value = my_value;        my_value1 = my_value1;        end;   end; run; -Urvish ... View more

Hi, If the Duplicates Value are in your master dataset then you need to execute your trnsaction dataset multiple times to apply the updates for all the observations in your master dataset...For that i have used SAS index to search for observation and execute the transaction dataset in such a way that updates will apply to all the observations of master dataset... Make sure that duplicates values are in MASTER DATASET not in TRANSACTION DATASET... proc sql;   create index name   on new_data(name); quit; data want(drop = ori_val);   do until(totobs);        set main_data(rename = (my_value = ori_val)) nobs = totobs;        set new_data key = name;        if _iorc_ NE 0 then do;       my_value = my_value;        end; end; run; -Urvish ... View more