Strange result from GAMLSS

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Strange result from GAMLSS

John Kerpel
Hi Folks!  Just started using the gamlss package and I tried a simple code
example (see below).  Why the negative sigma?

John


> y <- rt(100, df=1)> m1<-fitDist(y, type="realline")Warning messages:1: In MLE(ll3, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,  :
  possible convergence problem: optim gave code=1 false convergence
(8)2: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
 :
  possible convergence problem: optim gave code=1 false convergence
(8)3: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
 :
  possible convergence problem: optim gave code=1 false convergence
(8)4: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
 :
  possible convergence problem: optim gave code=1 false convergence
(8)5: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
 :
  possible convergence problem: optim gave code=1 false convergence (8)> > m1
Family:  c("TF", "t Family")
Fitting method: "nlminb"

Call:  gamlssML(y = y, family = DIST[i])

Mu Coefficients:
[1]  0.2376
Sigma Coefficients:
[1]  -0.09865
Nu Coefficients:
[1]  -0.2281

 Degrees of Freedom for the fit: 3 Residual Deg. of Freedom   97
Global Deviance:     545.069
            AIC:     551.069
            SBC:     558.885


Another draw:


> y <- rt(100, df=1)> m1<-fitDist(y, type="realline")Warning messages:1: In MLE(ll3, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,  :
  possible convergence problem: optim gave code=1 false convergence
(8)2: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
 :
  possible convergence problem: optim gave code=1 false convergence
(8)3: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
 :
  possible convergence problem: optim gave code=1 false convergence
(8)4: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
 :
  possible convergence problem: optim gave code=1 false convergence
(8)5: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
 :
  possible convergence problem: optim gave code=1 false convergence (8)> > m1
Family:  c("ST1", "Skew t (Azzalini type 1)")
Fitting method: "nlminb"

Call:  gamlssML(y = y, family = DIST[i])

Mu Coefficients:
[1]  -0.05249
Sigma Coefficients:
[1]  -0.1545
Nu Coefficients:
[1]  0.04445
Tau Coefficients:
[1]  -0.1352

 Degrees of Freedom for the fit: 4 Residual Deg. of Freedom   96
Global Deviance:     505.414
            AIC:     513.414
            SBC:     523.835



> sessionInfo()R version 2.15.1 (2012-06-22)
Platform: x86_64-pc-mingw32/x64 (64-bit)

locale:
[1] LC_COLLATE=English_United States.1252  LC_CTYPE=English_United
States.1252    LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C                           LC_TIME=English_United
States.1252

attached base packages:
[1] splines   stats     graphics  grDevices utils     datasets
methods   base

other attached packages:
[1] gamlss_4.1-8      gamlss.data_4.1-0 gamlss.dist_4.1-6 nlme_3.1-104
     MASS_7.3-21

loaded via a namespace (and not attached):
[1] grid_2.15.1      lattice_0.20-10  survival_2.36-14 tools_2.15.1

        [[alternative HTML version deleted]]

______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Reply | Threaded
Open this post in threaded view
|

Re: Strange result from GAMLSS

Ingmar Visser
Hi John,
I don't have access to R right now but do remember that some of the parametersin gamlss have log and/or logit link functions which would give the possibility for negative values.
Hth, Ingmar

Department of Psychology
University of Amsterdam

Op 11 Sep 2012 om 21:01 heeft John Kerpel <[hidden email]> het volgende geschreven:

> Hi Folks!  Just started using the gamlss package and I tried a simple code
> example (see below).  Why the negative sigma?
>
> John
>
>
>> y <- rt(100, df=1)> m1<-fitDist(y, type="realline")Warning messages:1: In MLE(ll3, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,  :
>  possible convergence problem: optim gave code=1 false convergence
> (8)2: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
> :
>  possible convergence problem: optim gave code=1 false convergence
> (8)3: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
> :
>  possible convergence problem: optim gave code=1 false convergence
> (8)4: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
> :
>  possible convergence problem: optim gave code=1 false convergence
> (8)5: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
> :
>  possible convergence problem: optim gave code=1 false convergence (8)> > m1
> Family:  c("TF", "t Family")
> Fitting method: "nlminb"
>
> Call:  gamlssML(y = y, family = DIST[i])
>
> Mu Coefficients:
> [1]  0.2376
> Sigma Coefficients:
> [1]  -0.09865
> Nu Coefficients:
> [1]  -0.2281
>
> Degrees of Freedom for the fit: 3 Residual Deg. of Freedom   97
> Global Deviance:     545.069
>            AIC:     551.069
>            SBC:     558.885
>
>
> Another draw:
>
>
>> y <- rt(100, df=1)> m1<-fitDist(y, type="realline")Warning messages:1: In MLE(ll3, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,  :
>  possible convergence problem: optim gave code=1 false convergence
> (8)2: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
> :
>  possible convergence problem: optim gave code=1 false convergence
> (8)3: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
> :
>  possible convergence problem: optim gave code=1 false convergence
> (8)4: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
> :
>  possible convergence problem: optim gave code=1 false convergence
> (8)5: In MLE(ll4, start = list(eta.mu = eta.mu, eta.sigma = eta.sigma,
> :
>  possible convergence problem: optim gave code=1 false convergence (8)> > m1
> Family:  c("ST1", "Skew t (Azzalini type 1)")
> Fitting method: "nlminb"
>
> Call:  gamlssML(y = y, family = DIST[i])
>
> Mu Coefficients:
> [1]  -0.05249
> Sigma Coefficients:
> [1]  -0.1545
> Nu Coefficients:
> [1]  0.04445
> Tau Coefficients:
> [1]  -0.1352
>
> Degrees of Freedom for the fit: 4 Residual Deg. of Freedom   96
> Global Deviance:     505.414
>            AIC:     513.414
>            SBC:     523.835
>
>
>
>> sessionInfo()R version 2.15.1 (2012-06-22)
> Platform: x86_64-pc-mingw32/x64 (64-bit)
>
> locale:
> [1] LC_COLLATE=English_United States.1252  LC_CTYPE=English_United
> States.1252    LC_MONETARY=English_United States.1252
> [4] LC_NUMERIC=C                           LC_TIME=English_United
> States.1252
>
> attached base packages:
> [1] splines   stats     graphics  grDevices utils     datasets
> methods   base
>
> other attached packages:
> [1] gamlss_4.1-8      gamlss.data_4.1-0 gamlss.dist_4.1-6 nlme_3.1-104
>     MASS_7.3-21
>
> loaded via a namespace (and not attached):
> [1] grid_2.15.1      lattice_0.20-10  survival_2.36-14 tools_2.15.1
>
>    [[alternative HTML version deleted]]
>
> ______________________________________________
> [hidden email] mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.