iterations and unique Id's

[Ksharp]

HOHO. you must be kidding with me .

data cust;
   input id :$2. cust:$3. date:date9.; 
   format date date9.; 
   cards; 
a1            c11  20May2013
a1            c11  15Jul2013
a1            c12  08Jan2014
a1            c11  25Apr2014
a1            c11  25May2014
a1            c11  25Jun2014
a1            c11  25Jul2014
a2            c21  20Jun2013  
a2            c23  15Jul2013
a2            c21  08Aug2013
a2            c24  25Apr2014
a3            c31  20Jun2013
a3            c32  15Jul2013
a3            c31  08Aug2013
a3            c32  25Feb2014
a3            c31  30Jul2014
;;;;
   run;

%macro extract(start=may2013,end=apr2014);
%let s=01&start ;
%let e=%sysfunc(intnx(month,"01&end."d ,0,e),date9.);
data temp(keep=id changed vname);
 set cust(where=(date between "&s"d and "&e"d));
 by id  notsorted ; 
 length vname $ 40;
 retain vname "&start._&end";
 if cust ne lag(cust) then changed+1;
 if first.id then changed=0;
 if last.id;
run;
proc append base=x data=temp force;run;
%mend extract;
options mprint symbolgen;
%extract(start=may2013,end=apr2014)
%extract(start=Jun2013,end=May2014)     
%extract(start=Jul2013,end=Jun2014)     
%extract(start=Aug2013,end=jul2014)


proc sort data=x;by id;run;
proc transpose data=x out=want(drop=_:) ;
by id;
id vname;
var changed;
run;

Xia Keshan

[stat_sas]

data have;
   input acct_nbr $ custid $ date : date9.;
   format date date9.;
_custid=input(compress(custid,,'kd'),best.);
mon=month(date);
month=date;
format month monyy7.;
cards;
a1            c11  20Jun2013
a1            c11  15Jul2013
a1            c12  08Aug2013
a1            c11  25Aug2013
a2            c21  20Jun2013
a2            c23  15Jul2013
a2            c21  08Aug2013
a2            c24  25Aug2013
a3            c31  20Jun2013
a3            c32  15Jul2013
a3            c31  08Aug2013
a3            c32  25Aug2013
a3            c31  30Aug2013
;;;;
   run;

data want (keep=acct_nbr month shift);
set have;
by acct_nbr mon _custid notsorted;
if first.acct_nbr and first.mon then shift=0;
if not first.mon and _custid-lag(_custid) ne 0 then shift+1;
if last.mon then output;
run;

proc tabulate data=want;
class acct_nbr month;
var shift;
table acct_nbr=' ',month=' '*shift=' '*sum=' '/box='acct_nbr';
run;