0

I have three vectors with sampled values, each from a specific probability distribution. (The vectors contain time series data of the growth rate of retail prices of specific goods)

%pearsrnd(mu,sigma,skew,kurt,m,n)
commodity1=pearsrnd(0.005,0.085,0.237,7.899,1,600);
commodity2=pearsrnd(0.003,0.040,0.280,5.630,1,600);
commodity3=pearsrnd(0.006,0.139,-0.207,4.209,1,600);

However, I want to add two correlation to the data. Specifically, the correlation between commodity1 and commodity3 should be 0.135, the correlation between commodity2 and commodity3 should be 0.115. How can I solve this most easily?

Mikhail_Sam
  • 10,602
  • 11
  • 66
  • 102
AJM
  • 17
  • 4
  • 1
    This question is probably more suited to http://stats.stackexchange.com/ Anyway, in general it's difficult to achieve a specified (marginal) distribution along with a set of correlations. As far as I know, only the Gaussian case lends itself easily to that – Luis Mendo May 15 '14 at 16:24
  • Thanks for your suggestion, I posted it there. Hope there is some sort of solution. – AJM May 15 '14 at 16:49

0 Answers0