proc sort data=have; by id city; run; data want(rename=(id2=id)); set have; by id city; retain id2; if first.id then id2=id; if first.city and ^first.id then id2 = id||city; drop id; run; ... View more

  proc sort data=have; by id city; run; data want; set have; by id city; retain id2; if first.id then id2=id; if first.city and ^first.id then id2 = id||city; run;   I hope this helps ... View more

data new; set old; string=repeat(strip(put(int(_n_/5) +1,$32.)),5); if mod(_n_,5) ==0 then string=repeat(strip(put(int(_n_/5) ,$32.)),5); run; Is this what you want? ... View more

Yes I agree you can use a hash table and you can use explicit loops based on point=, nobs usage. However a plain data step merge with a by statement would not do a many to many merge with the correct results, as in the definition of a many to many join. ... View more

There is no way you can do a many to many merge in a datastep using merge and by . so you have to resort to proc sql like Novinosrin mentioned below ... View more

data table_2a(keep=patient_age patient_age_last_follow_up Cum_dth_rate) table_2b(keep=patient age Cum_dth_rate) table_3a(keep=year patient Cum_dth_rate) table_3b (keep=age patient Cum_dth_rate); length patient $32.; array age_death_rates{100} (0.023, 0.023, 0.023, 0.023, 0.037, 0.037, 0.037, 0.037, 0.037, 0.063, 0.063, 0.063, 0.063, 0.063, 0.147, 0.147, 0.147, 0.147, 0.147, 0.147, 0.023, 0.023, 0.023, 0.023, 0.037, 0.037, 0.037, 0.037, 0.037, 0.063, 0.063, 0.063, 0.063, 0.063, 0.147, 0.147, 0.147, 0.147, 0.147, 0.147, 0.023, 0.023, 0.023, 0.023, 0.037, 0.037, 0.037, 0.037, 0.037, 0.063, 0.063, 0.063, 0.063, 0.063, 0.147, 0.147, 0.147, 0.147, 0.147, 0.147, 0.023, 0.023, 0.023, 0.023, 0.037, 0.037, 0.037, 0.037, 0.037, 0.063, 0.063, 0.063, 0.063, 0.063, 0.147, 0.147, 0.147, 0.147, 0.147, 0.147, 0.023, 0.023, 0.023, 0.023, 0.037, 0.037, 0.037, 0.037, 0.037, 0.063, 0.063, 0.063, 0.063, 0.063, 0.147, 0.147, 0.147, 0.147, 0.147, 0.147 ); input patient_age patient_age_last_follow_up; retain count 1; Cum_dth_rate=0; do age=patient_age to patient_age_last_follow_up-1; Cum_dth_rate = age_death_rates[age] + Cum_dth_rate ; put age_death_rates[age]=; end; output table_2a; Cum_dth_rate = 0; do age=patient_age to patient_age_last_follow_up; patient='patient_cnt'||strip(put(count,12.)); output table_2b; Cum_dth_rate = age_death_rates[age] + Cum_dth_rate ; end; Cum_dth_rate = 0; do age=patient_age+1 to patient_age+10; patient='patient_cnt'||strip(put(count,12.)); Cum_dth_rate = age_death_rates[age-1] + Cum_dth_rate ; year=age-patient_age; output table_3a; output table_3b; end; count=count+1; cards; 74 80 78 84 80 83 ;;; run; proc sort data=table_3b; by age; proc transpose data=table_3b out=table_3b_trans(drop=_name_); by age; id patient; var Cum_dth_rate; ; run; proc sort data=table_3a; by year; proc transpose data=table_3a out=table_3a_trans(drop=_name_); by year; id patient; var Cum_dth_rate; ; run; proc sort data=table_2b; by age; proc transpose data=table_2b out=table_2b_trans(drop=_name_); by age; id patient; var Cum_dth_rate; ; run; I created a random array of cumulative death rates by year. Please verify before you use it. ... View more

  proc means data = have nway missing noprint; class ID detailed_1  ; var _:; output out=want(drop=_type_ _freq_) max=; run;     ... View more

You can use a proc summary and do a max or proc means with a max and get the result.   proc means data = have nway missing noprint; class ID; var _:; output out=want(drop=_type_ _freq_) max=; run; ... View more

You are talking about stuff that system admins have nightmares handling on windows. A user can inadvertently permanently lock an excel file. ... View more