Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: AW: st: Weibull


From   Nikolaos Pandis <npandis@yahoo.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: AW: st: Weibull
Date   Sun, 8 Mar 2009 10:53:33 -0700 (PDT)

Dear Muhammad and Martin,
 
Your emails are greatly appreciated.
The objective is to create 4 different columns (variables) with  data conforming to the Weibull distribution ( timevar1, eventvar1/ timevar2, eventvar2) and then test to see whether there is a statistical difference between the two weibull variables regarding the events in conjunction with time . 
Then use exactly the same dataset,  count the events for the two different groups and see if there is a difference (chi2) when analysing them without taking in account the time information. The aim would be to show how the results of the analysis might differ when using the two different ways to analyse the same data.
 
I would like to generate weibull data with different parameters  and redo the above tests, and finally graph them.
 
Any assistance would be appreciated since this is unfamiliar territory for me.
 
Also the code you have posted says version 9.2-is this from an older Stata version? 
I am producing gamma distribution or weibull with the code you have created?
 
Thank you so much for taking the time to read and respond to my emails.
 
Best wishes,
 
Nikolaos

--- On Sun, 3/8/09, Martin Weiss <martin.weiss1@gmx.de> wrote:


From: Martin Weiss <martin.weiss1@gmx.de>
Subject: AW: st: Weibull
To: statalist@hsphsun2.harvard.edu
Date: Sunday, March 8, 2009, 4:49 PM



<> 

I think you can economize on lines here:


*************
capture program drop wsim
program define wsim
    version 9.2
    syntax [, obs(integer 1) lambda(real 0.8) gamma(real 0.4) ]
    drop _all
    set obs `obs'
    tempvar time event
    gen `time' = ((log(1-uniform()))/-`lambda')^(1/`gamma')
    summarize `time', mean
end
simulate time=r(mean), reps(1000): wsim, obs(100)
gen event= time< 5
*************



HTH
Martin


-----Ursprüngliche Nachricht-----
Von: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von Muhammad Riaz
Gesendet: Sonntag, 8. März 2009 15:45
An: statalist@hsphsun2.harvard.edu
Betreff: Re: st: Weibull

Hi Nikolaos,
You might have sorted out your simulation from weibull, I just managed to
write the codes given below: I thought it may help to thread it here  in
connection to your  query  for simulation from weibull and get further
advices from others.


capture program drop wsim
program define wsim, rclass
version 9.2
syntax [, obs(integer 1) lambda(real 0.8) gamma(real 0.4) ]
drop _all
set obs `obs'
tempvar time event
gen `time' = ((log(1-uniform()))/-`lambda')^(1/`gamma')
summarize `time'
return scalar mean = r(mean)

end
simulate time=r(mean) , reps(1000): wsim, obs(100)
gen event= time< 5

HTH

Muhammad Riaz   

--- On Mon, 3/2/09, Nikolaos Pandis <npandis@yahoo.com> wrote:

From: Nikolaos Pandis <npandis@yahoo.com>
Subject: Re: st: Weibull
To: statalist@hsphsun2.harvard.edu
Date: Monday, March 2, 2009, 9:48 PM

Dear Bobby,

Thank you very much for the posting. 

I have the previous version of the book on survival analysis with Stata that
you have coauthored with M. Cleves and W. Gould. Any sections in there that
will help me with the current problem?

Best wishes,

Nikolaos

--- On Mon, 3/2/09, Roberto G. Gutierrez, StataCorp <rgutierrez@stata.com>
wrote:


From: Roberto G. Gutierrez, StataCorp <rgutierrez@stata.com>
Subject: Re: st: Weibull
To: statalist@hsphsun2.harvard.edu
Date: Monday, March 2, 2009, 6:15 PM


Nikolaos Pandis <npandis@yahoo.com> asks:

> I would like to generate continuous data for two groups for which I would
> like to evaluate whether there is a difference between their means (I
could
> use use a t-test). I would like to be able to set the mean and sd of the
two
> groups. The data would represent bond failure strength (in MPa) of two
> different materials (the 2 groups).

> Additionally, I would like to generate data for the same experiment that
> follows the Weibull distribution and compare the two groups.

> It is of interest to see the results of the analysis on the data under the
> ttest analysis and under an analysis suitable for the Weibull
distribution.
> The objective is to show how different analyses of similar data might
point
> to different results, and the importance of selecting the correct
> distribution and analysis.

Nikolaos has already received some very good advice, to which I would only
add
that I have written the routine -gengammareg- available via  

   . findit gengammareg

-gengammareg- generates data from gamma regression as implemented by
-streg-,
but also includes as special cases the exponential, Weibull and lognormal
distributions; see -help gengammareg- after installing.

The two-sample test Nikolaos describes can be retooled as regression on a 
dummy variable, allowing the use of -gengammareg- here.

--Bobby
rgutierrez@stata.com
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/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/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/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/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/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index