SAS转置数据集

Question

I have a large transaction dataset need to do transpose. data I have:

id        prod
1            A
1            B
1            C
1            B
1            B
2            A
2            B
2            B
2            B
2            D

I need to transpose it to

id   PROD_1   PROD_2   PROD_3
1      A        B        C 
2      A        B        D

there are a lot of variables need to do this type of work. Any help is greatly appreciated. Really have no clue right now. Or if you have better idea to transform this information about prod into data set that will be able to analyze, please let me know.

Answer 1

You can transpose as many variables as you like in one data step. This will typically be much faster than doing the same thing with proc transpose:

data want;
  if 0 then set have; /*Keeps all columns in the original order*/
  array prods[5] $ prod1-prod5;
  do _n_ = 1 by 1 until(last.id);
    set have;
    by id;
    prods[_n_] = prod;
  run;
run;

Just add more arrays as necessary for each variable you want to transpose. This assumes that you only want to look at the same number of rows for each id - if you're not sure how many there are, you'll need to do an extra initial pass to find out how large you need to make the arrays.

This technique is known as a DOW-loop. Further reading: http://analytics.ncsu.edu/sesug/2010/BB13.Dorfman.pdf

Answer 2

If it is a large dataset, you need to think of efficiency. It needs to be sorted first (or indexed) on the ID variable. Also, ensure you only keep (process) the relevant variables:

proc transpose data=input(keep=id prod) out=output(drop=_name_) prefix=PROD_;
by id;
var prod;
run;