Creating Data Sets

Accepted Solution Solved
Reply
Occasional Contributor
Posts: 7
Accepted Solution

Creating Data Sets

I am trying to create different data sets on SAS with specifed variance/covariances on SAS university edition, but I am very unsure how to go about this. 

Thank you! 


Accepted Solutions
Solution
‎03-06-2016 08:36 AM
Super User
Posts: 19,058

Re: Creating Data Sets

Your trying to simulate data? If so read Rick Wicklins blog and/or book on simulating data in SAS. 

View solution in original post


All Replies
Solution
‎03-06-2016 08:36 AM
Super User
Posts: 19,058

Re: Creating Data Sets

Your trying to simulate data? If so read Rick Wicklins blog and/or book on simulating data in SAS. 

Occasional Contributor
Posts: 7

Re: Creating Data Sets

I have used Rick Wilkins' method to simulate multivariate normal data with a set covaraince matrix. I am ultimately trying to carry out mixed model analysis on this data. I am unsure how to split it up into groups, as it is already split into 'x1'...'x4'. This is the code I am using, if you happen to have any ideas?

 

proc iml;
Mean = {1, 2, 3, 4};
Cov = {0.5 0 0 0, 0 0.5 0 0, 0 0 0.5 0, 0 0 0 0.5};
N = 1000;
call randseed(123);
X = RandNormal(N, Mean, Cov);
SampleMean = mean(X);
SampleCov = cov(X);
varNames = "x1":"x4";

print SampleMean[colname=varNames],
SampleCov[colname=varNames rowname=VarNames];
create MVN from X[colname=varNames];
append from X;
close MVN;
quit;

SAS Super FREQ
Posts: 3,620

Re: Creating Data Sets

The example you show uses RANDNORMAL function to simulate correlated variables. Presumably you also want to generate correlated observations? The book Simulating Data with SAS covers mixed models in Chapter 12, pp. 230-242.  The article "Constructing common covariance structures" (from Chapter 10) shows some related computations.

Respected Advisor
Posts: 4,811

Re: Creating Data Sets

You will need to specify the distribution as well. Check out the RAND() function.

PG
☑ This topic is solved.

Need further help from the community? Please ask a new question.

Discussion stats
  • 4 replies
  • 391 views
  • 0 likes
  • 4 in conversation