Re: st: Re: Generating a sample from a two-Gaussian mixture

[Joerg Luedicke <joerg.luedicke@gmail.com>]

If I understand you correctly, you would like to generate two normal
mixture distributions with two components each, where the first
components (both with weight 0.4) of each of the two mixtures are
correlated with r = -0.1, and the two second components are correlated
with r = 0.1, right? Then this looks good to me. You can quickly check
it by typing:

bys p : sum x y // check M and SD of components
bys p : corr x y // check correlations

Joerg

On Fri, Mar 1, 2013 at 3:35 PM, Ali Hashemi <hashemi@vt.edu> wrote:
> Sorry, I just noticed a typo in the code. Here is the correct code:
>
> drawnorm x1 y1, n(1000) means(3,-1) sds(.2,.4) corr(1, -.1 \ -.1, 1)
> drawnorm x2 y2, n(1000) means(-1,1) sds(.3,.3) corr(1, .1 \ .1, 1)
> g p=uniform()<.4
> g x=p*x1+(1-p)*x2
> g y=p*y1+(1-p)*y2
>
> Thanks
> a.
>
> On Fri, Mar 1, 2013 at 2:36 PM, Ali Hashemi <hashemi@vt.edu> wrote:
>> Dear Statalist,
>>
>> I'm trying to generate a sample from the following mixture of two
>> normal distributions:
>>
>> 0.4*N ([3,-1];sigma=061) + 0.6*N ([-1,1];sigma=062)
>> where
>> sigma1=3D[0.2  -0.1 ; -0.1  0.4]
>> sigma2=3D[0.3   0.1 ;  0.1  0.3]
>>
>>
>> Here is my code:
>> drawnorm x1 y1, n(1000) means(3,-1) sds(.2,.4) corr(1, -.1 \ -.1, 1)
>> drawnorm x2 y2, n(1000) means(-1,1) sds(.3,.3) corr(1, .1 \ .1, 1)
>> g p=3Duniform()<.4
>> g x=3Dp*x1+(1-p)*x2
>> g y=3Dp*y1+(1-p)*y2
>>
>> I'm not sure whether this is the right way to do it. Is there a better
>> method for generating this sample? I would appreciate your help.
>>
>> Thanks
>> Ali Hashemi
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/