There are a couple of ways that you can order data within a table, using options in the PROC TABULATE statement itself, or within a CLASS statement. I think what you will want to do is use the CLASS statement. Within a CLASS statement you can order the data in the following ways (from the documentation): ORDER=DATA | FORMATTED | FREQ | UNFORMATTED                          DATA: orders values according to their order in the input data set.                                 FORMATTED: orders values by their ascending formatted values. This order depends on your operating environment.                                                          FREQ: orders values by descending  frequency count.                                 UNFORMATTED: orders values by their  unformatted values, which yields the same order as PROC SORT.                              So, I bet that if you sort the data in the order you want the dttm variables to appear, and create separate CLASS statement for the dttm variable (like shown), you will get what you need. Of course, this might mix up the order of other variables, so you might need to have separate CLASS statements for them, even using the PRELOADFMT option (which is a great feature!). One wonderful, but simple, resource for understanding PROC TABULATE is this paper: http://www2.sas.com/proceedings/sugi30/258-30.pdf. Good luck! proc sort data = dataln; by dttm; run; proc tabulate data = dataIn;      var catCount;      class dttm / order = data;      class studyID orderBy desc type;      table studyID*type*orderBy*desc*(style=Header),           catCount*dttm*max=' '; run; ... View more

Such question could simply be solved by this . I have a paper at SGF2015 about this skill . If you want to listen . Quad room, Apr 28, Peter.E will present it for me . data have;   length Study $5 Category $9;   infile datalines;   input study $ datetime : date7. category $ value;   format datetime date7.; return; datalines; x 01Jan13 Category1 10 x 01Jan13 Category2 20 x 01Jan13 Category3 30 x 01Jan13 Category4 40  x 07Jan13 Category1 20 x 07Jan13 Category3 40   x 14Jan13 Category2 27 y 03Jan13 Category1 15 y 04Jan13 Category1 25 y 01Jan13 Category3 35 y 03Jan13 Category4 45  y 09Jan13 Category1 25 y 09Jan13 Category3 45   y 14Jan13 Category2 29 ; run; proc sql; select distinct catt('have(where=(study="',study,'" and category="',category,'" and datetime=',datetime,')  rename=(value=_',put(datetime,date7. -l),'))')                     into : list separated by ' '   from have; quit; %put &list ; data want; merge &list ; by study category; drop datetime; run; Xia Keshan ... View more

It's for sure a good idea to specify the encoding explicitly. Some UTF8 encoded files don't have a BOM so telling SAS which encoding to use will help avoiding issues. ... View more

When I try to understand your question... I would say you want to have imagemaps defined on a webpage to generated PDF as hyperlinks. IT is not proc report but gives some techniques. SAS/STAT(R) 9.3 User's Guide (Using the Output Delivery System / Example 20.9 HTML Output with Graphics and Hyperlinks) ... View more

Note: You CAN use a DT prefix with datetime or time intervals but you don't have to for SECOND, MINUTE and HOUR. You do have to use the DT prefix for DAY and longer intervals (i.e. DTMONTH) to tell them apart from date intervals. Of course, you do have to use the corresponding date or datetime values for the other arguments. SAS (as of 9.3 at least) doesn't check for proper correspondence. PG ... View more

The closest I can come requires that the data be preprocessed. The following code does what I think you are looking for. However if your variable names or lengths of output are any longer than I'm showing then you will get another block of variables. I find that I don't like FILE PRINT ODS as you don't have much control, hence the keep variables and the order in the data set is the output order. The ODS Style is controlling fontsize. If a style with a larger font is used then the lines are longer and PDF will break the columns yet again. My example has an ID variable so you can connect the separate lines of the same observation. data junk;    do id = 1 to 20;       Array x x1-x48;       do j=1 to dim(x);          x = ranuni(237);       end;       output;    end;    drop j;    format x1-x50 f5.4; run;   ods pdf file="d:\data\part1.pdf" style=meadow startpage=no; data _null_;    set junk (obs=10 keep=id x1-x24);    file print ods;    put  id x1-x24; run; data _null_;    set junk (obs=10 keep=id x25-x48);    file print ods;    put  id x25-x48 ; run; ods pdf startpage=now; data _null_;    set junk (firstobs=11 obs=20 keep=id x1-x24);    file print ods;    put  id x1-x24; run; data _null_;    set junk (firstobs=11 obs=20keep=id x25-x48);    file print ods;    put  id x25-x48 ; run; ods pdf close; ... View more

Don't use ID statement.  PROC COMPARE will compare record by record as if you said ID _N_; sort of. If you are doing what I think you are, QCing ADaM data, I would use record level compare.  No ID statement. ... View more

Recommended reading from Cynthia ... View more

I think this would work : proc sql; create table tableC as select tableA.* from      tableA natural join      tableB(drop=sequenceId) order by tableA.sequenceId; PG ... View more

The first step creates a list of the dataset names. The second step reads and concatenates all those datasets in the list and creates a new variable called dataType to keep the name of the dataset that contributed each observation. The third step is a query that joins the Subjects dataset with the allDataTypes dataset by pairing observations with the same subjId. Hth. PG ... View more