Assigning alphanumeric values

[jprosenbaum8908]

Hi, I want to assign a value in the sequence {X1, Y1, X2, Y2}, starting with a random value from that sequence then cycle through the sequence over and over for the number of observations. For example:

if the random value from the sequence {X1, Y1, X2, Y2} is Y1, then the next value should be X2, then Y2, then X1, then Y1, then X2 and so on. 

Not sure how to do the alphanumeric assignment.

Thanks,

Josh

[Reeza · Accepted solution]

Set up a temporary array to hold your sequence. Then use a random distribution to get your first index and then increment from there, using the mod function to stay within the limits of your array.

data have;

do i=1 to 100;

output;

end;

run;

data want;

set have;

array seq(4) $ _temporary_ ("X1" "Y1" "X2" "Y2");

retain start;

if _n_=1 then start=floor(rand('uniform')*dim(seq))+1;

else start=mod(start,dim(seq))+1;

sequence=seq(start);

run;

[ballardw]

One way, possibly not the slickest:

data want;

     set have;

     retain pos;

     if _n_ = 1 then pos= mod(rand('table', 0.25, 0.25, 0.25, 0.25),4);

     else pos = mod(pos+1,4);

     select (pos);

          when(0) sequence='X1';

          when(1) sequence='Y1';

          when(2) sequence='X2';

          when(3) sequence='Y2';

          otherwise;

     end;

run;

You may want to drop the POS variable. Alternatively, skip the select and assignment to sequence and use a custom format on the POS variable (named as you like) that would display the X1, Y1 etc. That would actually be my choice.

[PGStats]

Simple task:

data have;

do i = 1 to 10; output; end;

run;

data want;

array values{4} $ _temporary_ ("X1", "Y1", "X2", "Y2");

if rndOffset=0 then rndOffset + ceil(dim(values)*rand("UNIFORM"));

set have;

rndValue = values{1 + mod(_n_ + rndOffset, dim(values))};

drop rndOffset;

run;

PG