[ADMB Users] glmmadmb problems

Thu Nov 21 01:18:10 PST 2013

Hi all,

My name is Daniel Moreno. I am working in forest regeneration and I am
trying to explain the factors which involves the regeneration in a
experiment composed by transects, plots and subplots. I have count data and
I would like to fit a negative binomial with a log link function. Below,
you can see the structure of the full data (excuse me for the Spanish code):

 str(todonava)
'data.frame':   170 obs. of  41 variables:
 $ COD            : int  9 10 11 12 13 14 15 16 17 18 ...
 $ Subplot        : Factor w/ 170 levels "1C5-3-1","1C5-3-2",..: 1 2 3 4 5
6 7 8 9 10 ...
 $ Plot           : Factor w/ 43 levels "1C5-3","1C5-4",..: 1 1 1 1 2 2 2 2
3 3 ...
 $ Transecto      : Factor w/ 5 levels "1C5","2A5","2B5",..: 1 1 1 1 1 1 1
1 1 1 ...
 $ Seedlings      : int  9 59 5 92 8 11 4 4 22 33 ...
 $ isf            : num  0.852 0.848 0.855 0.862 0.82 ...
 $ dsf            : num  0.812 0.785 0.807 0.835 0.812 ...
 $ gsf            : num  0.817 0.793 0.813 0.838 0.813 ...
 $ altitud        : int  1730 1730 1730 1730 1730 1730 1730 1730 1730 1730
...
 $ pendgrados     : num  14 11 12 16 16 15 15 17 19 11 ...
 $ pend100        : num  25 20 22 29 29 27 28 32 35 20 ...
 $ numfoto        : int  9 10 11 12 13 14 15 16 17 18 ...
 $ ph             : num  4.67 4.67 4.67 4.67 8.45 8.45 8.45 8.45 4.76 4.76
...
 $ conductividad  : num  0.08 0.08 0.08 0.08 0.2 0.2 0.2 0.2 0.08 0.08 ...
 $ arena          : num  71 71 71 71 58 ...
 $ limo           : num  11 11 11 11 12.6 ...
 $ arcilla        : num  18.4 18.4 18.4 18.4 29.4 ...
 $ MO             : num  13.9 13.9 13.9 13.9 17.4 ...
 $ potasio        : int  134 134 134 134 182 182 182 182 180 180 ...
 $ fosforo        : num  11.5 11.5 11.5 11.5 13.5 13.5 13.5 13.5 8.8 8.8 ...

Some variables contains NA values (e.g. fosforo (fosforo=phosphorus)).
Below, it is possible to read my R code (to fit the best model and avoid
overparametization I use a stepwise model selection) and some Errors:

> todo2<-glmmadmb(Seedlings~Transecto + fosforo + (1|Plot), na.omit,
data=todonava, zeroInflation=FALSE,   family="nbinom",  link="log")

Error in II[, ii] = II[, ii] + REmat$codes[[i]] :
  number of items to replace is not a multiple of replacement length
In addition: Warning messages:
1: In glmmadmb(Seedlings ~ Transecto + fosforo + (1 | Plot), data =
todonava,  :
  NAs removed in constructing fixed-effect model frame: you should probably
remove them manually, e.g. with na.omit()
2: In II[, ii] + REmat$codes[[i]] :
  longer object length is not a multiple of shorter object length

It could be a NA problem. Then, I removed NA values:

navana<-na.omit(todonava)

 str(navana)
'data.frame':   0 obs. of  41 variables:
 $ COD            : int
 $ Subplot        : Factor w/ 170 levels "1C5-3-1","1C5-3-2",..:
 $ Plot           : Factor w/ 43 levels "1C5-3","1C5-4",..:
 $ Transecto      : Factor w/ 5 levels "1C5","2A5","2B5",..:
 $ Seedlings      : int
 $ isf            : num
 $ dsf            : num
 $ gsf            : num
 $ altitud        : int
 $ pendgrados     : num
 $ pend100        : num
 $ numfoto        : int
 $ ph             : num
 $ conductividad  : num
 $ arena          : num
 $ limo           : num
 $ arcilla        : num
 $ MO             : num
 $ potasio        : int
 $ fosforo        : num
 $ magnesio       : num
 $ calcio         : num
 $ sodio          : num
 $ texturaISSS    : Factor w/ 5 levels " arcillo arenoso",..:
 $ F25            : logi
 $ lnca           : num
 $ lnconductividad: num
 $ lnfosforo      : num
 $ gsfcuadrado    : num
 $ lnMg           : num
 $ lnpotasio      : num
 $ raizpotasio    : num
 $ lnsodio        : num
 $ ipot           : num
 $ ifn            : num
 $ trees          : int
 $ stump          : int
 $ perctrees      : num
 $ percstumps     : num
 $ pedre          : num
 $ Lnpedre        : num
 - attr(*, "na.action")=Class 'omit'  Named int [1:170] 1 2 3 4 5 6 7 8 9
10 ...
  .. ..- attr(*, "names")= chr [1:170] "1" "2" "3" "4" ...

Then, I run again the model with the new data frame:

> todo3<-glmmadmb(Seedlings~Transecto + fosforo + (1|Plot), na.omit,
data=navana, zeroInflation=FALSE,   family="nbinom",  link="log")
Error in `contrasts<-`(`*tmp*`, value = contr.funs[1 + isOF[nn]]) :
  contrasts can be applied only to factors with 2 or more levels

I am new with R (till now, I worked with SAS). Could you help me to solve
this problems)

Thanks you all,

Daniel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.admb-project.org/pipermail/users/attachments/20131121/d936e504/attachment.html>