Start a process when start_variable=1

[hhchenfx]

Hi Everyone,

I have a start variable having value 1.

When value =1, I want to run the below 2 data steps.

Of course, I can run the whole process on every single record but it is not optimal.

I appreciate it if you could help me.

HHC

data have;
input start order value;
datalines;
0 1 5
0 2 2
0 3 4
0 4 4
0 5 5
0 6 2
0 7 7
0 8 1
0 9 13
0 1 10 5
0 1 56
0 2 26
0 3 46
0 4 46
0 5 18
0 6 58
0 7 8
0 8 2
0 9 100
1 10 5
;
run;

IF START=1 THEN DO THE FOLLOWING PROCESS

data want;

set have;
drop aa1-aa10;
array a{0:9} aa1-aa10;
retain aa:;
pct_60=largest((10-10*60/100+1),of a{*});
pct_70=largest((10-10*70/100+1),of a{*});
pct_80=largest((10-10*80/100+1),of a{*});
pct_90=largest((10-10*90/100+1),of a{*});
a{mod(_n_,10)}=value;
run;

data want; set want;
drop pct_25-pct_100 ;
if value>=pct_90 then ROC_rank=5;
else if value>=pct_80 then ROC_rank=4;
else if value>=pct_70 then ROC_rank=3;
else if value>=pct_60 then ROC_rank=1.5;
else ROC_rank=0.1;
run;

[PGStats]

Your example code generates an error: it is interrupted because of looping. What is it intended to do?

[hhchenfx]

Sorry for not deleting the IF statement.

Here is the clean one.

Thank you for helping, PGStats.

HHC

data have;
input start order value;
datalines;
0 1 5
0 2 2
0 3 4
0 4 4
0 5 5
0 6 2
0 7 7
0 8 1
0 9 13
0 1 10 5
0 1 56
0 2 26
0 3 46
0 4 46
0 5 18
0 6 58
0 7 8
0 8 2
0 9 100
1 10 5
;
run;

*IF START=1 THEN DO THE FOLLOWING PROCESS;

data want;

set have;
drop aa1-aa10;
array a{0:9} aa1-aa10;
retain aa:;
pct_60=largest((10-10*60/100+1),of a{*});
pct_70=largest((10-10*70/100+1),of a{*});
pct_80=largest((10-10*80/100+1),of a{*});
pct_90=largest((10-10*90/100+1),of a{*});
a{mod(_n_,10)}=value;
run;

data want; set want;
drop pct_25-pct_100 ;
if value>=pct_90 then ROC_rank=5;
else if value>=pct_80 then ROC_rank=4;
else if value>=pct_70 then ROC_rank=3;
else if value>=pct_60 then ROC_rank=1.5;
else ROC_rank=0.1;
run;

[PGStats · Accepted solution]

I'm not certain I understand fully. It seems like you want to compare value with the preceeding 10 values and set ROC_rank based on the ranking of value relative to the set of 10. Furthermore, you want output only for cases where start=1. The role of variable order is undefined. This would do that:

data want;
set have;
array a{0:9} _temporary_;
if start and nmiss(of a{*})=0 then do;
    if      value>=largest(1,of a{*}) then ROC_rank=5;
    else if value>=largest(2,of a{*}) then ROC_rank=4;
    else if value>=largest(3,of a{*}) then ROC_rank=3;
    else if value>=largest(4,of a{*}) then ROC_rank=1.5;
    else                                   ROC_rank=0.1;
    output;
    end;
a{mod(_n_,10)}=value;
run; 

[hhchenfx]

Hi PGStats,

Your code work perfectly in this case. And yes, I only need to output record with start=1.

However, I did not foresee that it can be done in only 1 step.

And my code to run when start=1 involve few other steps.

So it is the best for me if I can put the MAIN analysis section within some kind IF as below.

It will help me alot for other applicaiton as well.

Thank you so much.

HHC

IF start=1 then do;

data want;
set have;
drop aa1-aa10;
array a{0:9} aa1-aa10;
retain aa:;
pct_60=largest((10-10*60/100+1),of a{*});
pct_70=largest((10-10*70/100+1),of a{*});
pct_80=largest((10-10*80/100+1),of a{*});
pct_90=largest((10-10*90/100+1),of a{*});
a{mod(_n_,10)}=value;
run;

END;

data want; set want;
drop pct_25-pct_100 ;
if value>=pct_90 then ROC_rank=5;
else if value>=pct_80 then ROC_rank=4;
else if value>=pct_70 then ROC_rank=3;
else if value>=pct_60 then ROC_rank=1.5;
else ROC_rank=0.1;
run;