topic cumulative numbering of type using hash in SAS Programming

cumulative numbering of type using hash

km0927 — Tue, 30 Oct 2018 01:07:01 GMT

hello. i would like to ask about cumulative numbering of clinic type.

i've been asked about various type of cumulative calculation in this community. i'm so thankful for all your help so far.

however, 'hash' command is very difficult to apply in new question. so, i asked one more question.

i want to calculative type of clinic cumulatively. raw data is here....

data have;
input id date clinic ;
cards;
1 20020408 12135
1 20020715 12135
1 20030124 15109
1 20030308 75423
1 20030505 75423
1 20030607 75423
1 20030815 12135
2 20020615 25875
2 20020717 25875
2 20020824 55415
2 20020907 25875
2 20021215 25875
2 20030108 25875

and i want to make this raw data like this....

data want;
input id date clinic type;
cards;
1 20020408 12135 1
1 20020715 12135 1
1 20030124 15109 2
1 20030308 75423 3
1 20030505 75423 3
1 20030607 75423 3
1 20030815 12135 3
2 20020615 25875 1
2 20020717 25875 1
2 20020824 55415 2
2 20020907 25875 2
2 20021215 25875 2
2 20030108 25875 2

as you can see in 7th row, the numbering of type is still same, because clinic '12135' is not new type of clinic for 'id 1'

but in 10th row, 1 was added to the numbering of type, because clinic '55415' is new type of clinic for 'id 2'

i tried this code, but the results didn't worked out.

data want;
set have;
by id;
if _n_=1 the do;
declare hash h ();
h.definekey('clinic');
h.definedata('type');
h.definedone();
end;

if first.id then h.clear();
if h.find()=0 then do;
type=1;
h.replace();
end;

if h.find()^=0 then do;
type=type+1;
h.replace();
run;

i really appreciate for all your help.

Re: cumulative numbering of type using hash

ChrisNZ — Tue, 30 Oct 2018 01:28:57 GMT

Like this?


data WANT;
  length LIST $32000;
  retain LIST ;
  drop LIST;
  set HAVE; 
  by ID;
  if first.ID then do;
    LIST=cat(CLINIC);
    TYPE=1;
  end;
  else if ^findw(LIST,catt(CLINIC)) then do;
    TYPE+1;
    LIST=catx(' ',LIST,CLINIC);
  end;
run;

Hash tables are not the universal answer.

id	date	clinic	TYPE
1	20020408	12135	1
1	20020715	12135	1
1	20030124	15109	2
1	20030308	75423	3
1	20030505	75423	3
1	20030607	75423	3
1	20030815	12135	3
2	20020615	25875	1
2	20020717	25875	1
2	20020824	55415	2
2	20020907	25875	2
2	20021215	25875	2
2	20030108	25875	2

Re: cumulative numbering of type using hash

novinosrin — Tue, 30 Oct 2018 03:23:03 GMT


data have;
input id date :yymmdd10. clinic ;
format date yymmdd10.;
cards;
1 20020408 12135
1 20020715 12135
1 20030124 15109
1 20030308 75423
1 20030505 75423
1 20030607 75423
1 20030815 12135
2 20020615 25875
2 20020717 25875
2 20020824 55415
2 20020907 25875
2 20021215 25875
2 20030108 25875
;

data want;
if 0 then set have;
array t(999) _temporary_;
call missing(of t(*),type);
do until(last.id);
set have;
by id ;
if clinic not in t then do;
type+1;
t(type)=clinic;
end;
output;
end;
run;