Contributed by Rick Wicklin, author of Statistical Programming with SAS/IML Software
This year at SAS Global Forum 2011, I am presenting a statistical tutorial, "Data Simulation for Evaluating Statistical Methods in SAS". In this course, I show how to create data with known properties (such as skewed or heavy-tailed) and how to use that data to understand when using a particular statistic is valid or invalid. For example, can you use the t test for data that are highly skewed?
During my presentation, I'll be sharing some tips and techniques from my book, Statistical Programming with SAS/IML Software. The following tip is particularly important for efficiently using SAS/IML software to generate samples.
Tip:
To generate N random samples from a specified distribution, allocate a vector of length N and call the RANDGEN subroutine.
For example, to generate 1,000 random numbers from the standard normal distribution, you can use the following statements:
proc iml; x = j(1000, 1); /** allocate 1000 x 1 vector **/ call randgen(x, "Normal"); /** sample from random normal **/
My book has more than 150 other tips and techniques that can help you use SAS/IML software more efficiently for data analysis, sampling and simulation, and writing statistical analyses.
Do you have any questions for Rick? Submit a comment here or look for Rick in the SAS/IML and SAS/IML Studio discussion forum. You can also find additional tips on Rick's popular blog The DO Loop