Complete missing data in big dataset

[Freddy662]

Dear all,

I currently have a huge dataset which looks like this:

Value 1	value 2	Value 3	Value 4	Value 5
OK	OK	NAME1	.	OK
OK	OK	NAME2	Value2	OK
OK	OK		Value1	OK
OK	OK	NAME2	.	OK
OK	OK	NAME3	Value3	OK
OK	OK	NAME3	Value3	OK
OK	OK	NAME1	Value1	OK
OK	OK	NAME2	.	OK
...	...	...	...	...
OK	OK	NAME3	Value3	OK

In the end I would like to have something like this:

Value 1	value 2	Value 3	Value 4	Value 5
OK	OK	NAME1	Value1	OK
OK	OK	NAME2	Value2	OK
OK	OK		Value1	OK
OK	OK	NAME2	Value2	OK
OK	OK	NAME3	Value3	OK
OK	OK	NAME3	Value3	OK
OK	OK	NAME1	Value1	OK
OK	OK	NAME2	Value2	OK
...	...	...	...	...
OK	OK	NAME3	Value3	OK

I'm currently struggling with the code for this:

I have already the following datastep:

data &outlib..&outdsn.;

  Set &outlib..&outdsn;

if missing (value_4) and not missing (value_3) then

  do i=1 %to &vars_N.;

  do until (not missing(value_4));

but then nothing works further.

Can anyone help me out?

Thanks a lot in advance.

Frederik

[RW9]

Post test data in the form of a datastep in a code window!!

As such I am not typing that in.  Why should the third row be used to populate the first one?  I can see no logical reason.  Post something which actually reflects the problem.  Now at a guess, what you can do is to first create a distinct list of Name + Value pairs which are not missing, then merge that back onto the data e.g:

proc sort data=have out=inter (rename=(value4=rep)) nodupkey;
  by value3 value4;
  where value3 ne "" and value4 ne "";
run;

data want;
  merge have inter;
  by value3;
  if value4="" then value4=rep;
run;