Please help me...

[Cello23]

Hello everybody ,

i have this dataset:

I would like to have this result (if IDCAB  is blank must take the value of the previous cell):

I use this query:
data TABLE_1;

set  TABLE_1;

n=_n_;

if missing(IDCAB) then do;

do until (not missing(IDCAB));

n=n-1;

set TABLE_1(keep=IDCAB) point=n;

end;

end;

run;

it works, but with "big dataset" (such as 600,000 records) it crashes.

Please, Can you suggest an alternative?

Thank you.

[ieva · Accepted solution]

Try:

data table_2 (drop=idcab rename=idcab_mod=idcab);

set table_1;

retain idcab_mod ;

if idcab ne '' then idcab_mod =idcab;

run;

[Cello23]

YES!!!!!

" i love you IEVA" !!!  :smileylaugh:

thank you!!!   

[Cello23]

Sorry...!!!

I have another similar question...

i have this dataset:

I would like to have this result (if IDCAB  is blank and AID is equal to the previous cell,

IDCAB is  must take the value of the previous cell, If this is not true the cell should be blank):

This dont' work...

data TABLE_1(drop=IDCAB AID  rename=IDCAB_MOD=IDCAB rename=AID_MOD=AID);

set TABLE_2;

retain IDCAB_MOD AID_MOD ;

if (IDCAB ne '' AND AID_MOD=AID) then IDCAB_MOD =IDCAB;

run;

Please help me.

Thank you.

[art297]

For the cells that you want missing, the aid numbers are the same in your sample data, but different in your desired output.  Which is correct?

[ieva]

This could be the solution you are looking for:

proc sort data=table_1;

by aid descending idcab;

run;

data table_2 (drop=idcab rename=idcab_mod=idcab);

set table_1;

by aid descending idcab;

retain  idcab_mod ;

if first.aid  then idcab_mod =idcab;

run;

[Linlin]

data have;

  infile cards missover;

  input aid $ idcab $;

cards;

1 ad

1 bb

1

2 fv

2

2

2 cv

;

run;

data want (drop=idcab rename=idcab_mod=idcab);

set have;

by aid;

retain idcab_mod ;

if (first.aid or idcab ne '' )then idcab_mod =idcab;

run;

[art297]

Presuming that your output file contained the correct values, and that your file is already in the order shown in your example, then the following would do (I think) what you want:

data table_2 (drop=hold:);

  set table_1;

  by aid notsorted;

  retain hold_idcab;

  if first.aid then hold_idcab=idcab;

  else idcab=hold_idcab;

run;

[Cello23]

sorry for my mistake, output file container the correct values.

your support is really precious !!!!

Thanks to all!!!

[Linlin]

Dear  Art,

I ran your code. It did not have the correct outcome for the last observation.

data table_1;

  infile cards missover;

  input aid $ idcab $;

cards;

1 ad

1 bb

1

2 fv

2

2

2 cv

;

run;

data table_2 (drop=hold:);

  set table_1;

  by aid notsorted;

  retain hold_idcab;

  if first.aid then hold_idcab=idcab;

  else idcab=hold_idcab;

run;

proc print;

run;

                                      Obs    aid    idcab

                                        1      1      ad

                                        2      1      ad

                                        3      1      ad

                                        4      2      fv

                                        5      2      fv

                                        6      2      fv

                                        7      2      fv

[art297]

The OP didn't have such a condition in the example data, thus I decided that it would be presumptive to include one.  Obviously, if the OP wanted non-missing values were to be left alone, the decision would also have to be made regarding whether such new values should be assigned to later repititions as well.

Soapbox: Unfortunately, programmers often write code to cover conditions that are outside of the defined requirements, rather than simply mentioning the possible discrepancy(ies) and asking how it/they should be handled.

.