I have an existing data set ("test" in my example below) where indexing and sorting are very costly tasks in terms of time due to size. I am attempting to add a column (that exists on "test2" below) to that existing data set without losing the sorting and indexing that is done on them. A demonstration of the issue below. First I set up the problem by creating the analogous data.   data test; input c1 c2 c3 c4; datalines; 1 3 5 2 3 5 1 7 2 6 8 9 1 2 7 3 6 1 9 1 9 1 3 6 4 1 1 7 ; run; proc sort data=test; by c1 c2 c3; run; Proc Datasets library = work nolist; modify test; index create comp=(c1 c2); run; /*Verify Index and Sort*/ proc contents data=test varnum; data test2; input c1 c2 c3 newc $; datalines; 1 3 5 r 3 5 1 f 2 6 8 s 1 2 7 q 6 1 9 q 9 1 3 w 4 1 1 i ; run; proc sort data=test2; by c1 c2 c3; run; /*Verify Sort*/ proc contents data=test2 varnum; Notice that test is both sorted and indexed, and test2 is sorted. When I do a normal joining method like below it loses all the metadata around the indexing and sorting for test. How can I get the result below but where I would retain the sorting and indexing?   proc sql; create table test as select a.*, b.newc from test a left join test2 b on (a.c1=b.c1) And (a.c2=b.c2) And (a.c3=b.c3) ; quit; I have been attempting to use some version of proc sql's alter table, but am not sure how to actually add the column. I seem to be able to create an empty column, but not populate it with the data from test2. I would also be very interested in any data step solutions to this issue.   ... View more

I have a data set such as below:     data test; input c1 c2 c3 c4; datalines; 1 3 5 2 3 5 1 7 2 6 8 9 1 2 7 3 6 1 9 1 9 1 3 6 4 1 1 7 ; run; I have indexed it as follows:       data test2 (index=(c1 c2 )); set test; run; Suppose I want to index on an additional column at a later time, is there a way to add an index for c3, without deleting or re-indexing the columns c1 and c2? For my actual purposes, re-indexing columns c1 and c2 is very expensive as it is a very large data set. Obviously if you do the below you lose your indexes to 1 and 2:     data test2 (index=(c3 )); set test2; run; And if you do the following you re-index over all 3 columns (very computationally expensive): data test2 (index=(c1 c2 c3 )); set test; run; Is there anyway to get the results of indexing over all 3 columns, without having to redo the indexes on 1 and 2?       ... View more

We have a large data set that we are maintaining on a monthly basis. In order to make it more efficient for people on our team to use we are attempting to index it on a few different variables, which initially took quite a long time. I'm wondering if there is a way to update an existing index rather than re-indexing the entire data set every time we append to the data? My thinking is that if we are able to just take the new values in the set and add those to the index we could save significantly on time, but I'm not sure if the syntax exists to do that. I am creating my index with the below code.     data test2 (index=(c1 c2 c3 c4 c5 c6)); set test; run;       ... View more