Much the same really, just one datastep rather than 2. proc sort data=testdata;   by var1 date; run; data want (drop=lstdate);   set testdata;   retain lstdate;   by var1;   if first.var1 then lstdate=date;   if date in ('28sep2014'd,'28dec2014'd,'29mar2015'd,'28jun2015'd) and lag(var1)=var1 then do;     days_between=date - lstdate;     output;   end;   lstdate=date; run; ... View more

Code: Program data have;    infile cards truncover ;    input  url $200.;    cards; http://www.mywebsite.com/family/index.jsp?categoryId=61765546&cp=1766205&ab=en_US_MLP_SLOT_1_S1_SHOP http://www.mywebsite.com/shop/index.jsp?categoryId=62593996&AB=en_US_HP_S2_Men_slot_1_S2_ShopNow http://www.mywebsite.com/family/index.jsp?categoryId=62593846&cp=1766205&ab=ln_men_cs_thetrend:tropicalprints http://www.mywebsite.com/family/index.jsp?categoryId=62594006&cp=1766205&AB=en_US_MLP_P_slot_10_S1_ShopNow http://www.mywebsite.com/family/index.jsp?categoryId=56906456 http://www.mywebsite.com/shop/index.jsp?categoryId=62593866&cp=1766205&ab=ln_men_cs_theshirt http://www.mywebsite.com/family/index.jsp?categoryId=62593876&cp=1795710&AB=en_US_MLP__slot_2_S1_ShopNow http://www.mywebsite.com/shop/index.jsp?categoryId=57155616 http://www.mywebsite.com/shop/index.jsp?categoryId=1766618&ab=tn_women_golfandTennis&cp=17666 ; data want; set have; pid= prxparse('/(?<=categoryId=)\d+/i'); call prxsubstr(pid, url, position, length); if position ne 0 then do; match = substr(url, position, length); end; drop pid position length; run; ... View more