03-09-2012 12:13 PM
I mostly do my claculations in Proc Sql but it takes some good amount of steps.I came to know that we can use Proc Tabulate.I just want to know which is the better way and which takes less time if i have huge data.Any suggestions will be appreciated.
03-09-2012 01:46 PM
Depends on what you're calculating.
If you need distinct counts of something SQL will give that you more easily, perhaps HASH as well, but I'm not familiar with that.
Otherwise tabulate is very convenient, but proc means and summary also do very nice summaries. You can also look into the ODS tables to get more customizable output.
03-09-2012 03:33 PM
I generally think of SQL for data manipulation and TABULATE for data presentation.
Suppose I want to display some test results with nested leveles of display such as a region and subregions for rows and age-group and gender for columns and cells to represent mean and stand deviation of test scores.
Tabulate will do that quite nicely and present it in a pretty nice looking table. Adding in the ability to group and display data using MULTILABLE formats and/or forcing all formatted values of variables to appear if desired it gets pretty powerful.
You can get output from tabulate but in my experience it is an odd day when I need to.