topic Re: Index options for those digits that doesn't follow the defined pattern in SAS Enterprise Guide

Index options for those digits that doesn't follow the defined pattern

claudiopcjr — Tue, 18 Feb 2020 02:44:15 GMT

Hi everyone, I'm working with informations about customers along 136 months.

What I'm working on is an specific information about those customers that express a situation that can from a month to another. With this data in hands I'm building a report to specify the duration (in months) of each situation.

For example, concatenating all these situation for an hypothetic customer we would have:

customer_historic='00000666668888888888880000000000000000777777777777777777770000000006666666'

At first I could say that on the first five months the situation was 0 and lasted for 5 months.

And since the 6th month the situation began to be 6, and it lasted for 5 months as well.

And then, from the 11th month it started to be 8, and it lasted for 8 months

.....

But, using formulas, I just can identify the beginning of this sequence, which could be done using:

FIRST_NUMB_SEQUENCE = INDEX(customer_historic,'0');

Which the result would be FIRST_NUMB_SEQUENCE = '1'.

Does anyone know a way to identify when, along the sequence, the digit starts to be different from 6 ? Is possible to do that using INDEX function ?

Or an alternative way.

Thanks in advance.

Re: Index options for those digits that doesn't follow the defined pattern

mkeintz — Tue, 18 Feb 2020 03:07:33 GMT

Many suggestions might be made, but why not show us what the resulting data set would look like? That will help narrow down the relevant answers. For instance, if a customer's history has, say 7 "regimes", do you want seven output records (and what would they look like)? If some other structure, what would it look like?

Re: Index options for those digits that doesn't follow the defined pattern

Patrick — Tue, 18 Feb 2020 09:22:43 GMT

Here one way of doing this. Code based on example from docu here.

data have;
  id=1;
  customer_historic='00000666668888888888880000000000000000777777777777777777770000000006666666';
  output;
  id=2;
  call missing(customer_historic);
  output;
  stop;
run;

data want(drop=_:);
  set have;
  length value $1;
  _exprid = prxparse('/(\d)\1*/');
  _start = 1;
  _stop = lengthn(customer_historic);

  if _stop=0 then output;
  else
    do;
    /* Use PRXNEXT to find the first instance of the pattern, */
    /* then use DO WHILE to find all further instances.       */
    /* PRXNEXT changes the _start parameter so that searching  */
    /* begins again after the last match.                     */
    call prxnext(_exprid, _start, _stop, customer_historic, first_numb_sequence , _len);

    do while (first_numb_sequence  > 0);
      value = substrn(customer_historic, first_numb_sequence , 1);
      last_numb_sequence=first_numb_sequence+_len-1;
      output;
      call prxnext(_exprid, _start, _stop, customer_historic, first_numb_sequence , _len);
    end;
  end;
run;

proc print;
run;

Re: Index options for those digits that doesn't follow the defined pattern

claudiopcjr — Tue, 03 Mar 2020 19:47:23 GMT