Hello everyone,
I fitted a GLMM model on count data of the number of fruits produced by
açaí palms in the Amazon using the glmmTMB package. The model is negative
binomial with zero inflation, a random structure that controls for repeated
measurements on the same palms in 2 consecutive years ("number" variable)
and a spatial structure of nested plots within transects within blocks. The
fixed part contains the year of measurement "ano_medida", the soil height
diameter of each plant das_mm and other variables linked to the hydrology
and forest structure of each plot. The variables in the fixed part were
z-standardized to stabilize the model and deal with overdispersion. The
model was able to deal with an overdispersion problem and a high proportion
of structural zeros. The selection was made first for the zero-inflation
part and then for the conditional part.
I would like to graphically illustrate the relationship of the non zero
inflated part between n_fruits and das_mm conditional on the results found
in the model. I tried to do this using the visreg function but I get a
repeated error:
> visreg(modelo16_REML, "das_mm", type = "contrast", data =
dados_transformados)
Error in eval(predvars, data, env) :
objeto 'dif_alt_dossel_m' não encontrado
I tried to apply a solution I found in the visreg FAQ, but it didn't work:
fit$gam does not include the call (i.e., fit$gam$call is NULL), which means
visreg won’t be able to find the data:
visreg <https://pbreheny.github.io/visreg/reference/visreg.html>(fit$gam,
'x')
# Error in FUN(X[[i]], ...): object 'y' not found
So you have to include it manually:
fit$gam$data <- Data
visreg <https://pbreheny.github.io/visreg/reference/visreg.html>(fit$gam,
'x')
modelo16_REML$data = dados_transformados
However visreg did not work in this way also.
Can anyone tell me how to illustrate this relationship? I'm only referring
to the conditional part of the model, because I'm not going to illustrate
the zeros inflation part, it was only used to control the zeros inflation
part and it was illustrated separately through a logistic model, for which
visreg worked well.
Thanks in advance for any ideas,
Alexandre
modelo16_REML <-
glmmTMB(n_frutos ~
ano_medida + das_mm +
(1|numero) + (1|bloco/transecto/parcela),
ziformula = ~ ano_medida +
das_mm + dif_alt_dossel_m + elevation_m + hand_m +
basal_area_m2 + tree_density +
das_mm:dif_alt_dossel_m + elevation_m:hand_m +
(1|numero) + (1|bloco/transecto/parcela),
family = nbinom1,
data = dados_transformados, REML=TRUE)
> summary(modelo16_REML)
Family: nbinom1 ( log )
Formula:
n_frutos ~ ano_medida + das_mm + (1 | numero) + (1 |
bloco/transecto/parcela)
Zero inflation:
~ano_medida + das_mm + dif_alt_dossel_m + elevation_m + hand_m +
basal_area_m2 + tree_density + das_mm:dif_alt_dossel_m +
elevation_m:hand_m + (1 | numero) + (1 | bloco/transecto/parcela)
Data: dados_transformados
AIC BIC logLik deviance df.resid
22121.3 22255.7 -11036.7 22073.3 1975
Random effects:
Conditional model:
Groups Name Variance Std.Dev.
numero (Intercept) 0.019859 0.14092
parcela:transecto:bloco (Intercept) 0.006741 0.08211
transecto:bloco (Intercept) 0.015453 0.12431
bloco (Intercept) 0.020497 0.14317
Number of obs: 1995, groups:
numero, 761; parcela:transecto:bloco, 146; transecto:bloco, 39; bloco, 10
Zero-inflation model:
Groups Name Variance Std.Dev.
numero (Intercept) 1.323e+00 1.1500780
parcela:transecto:bloco (Intercept) 4.588e-01 0.6773243
transecto:bloco (Intercept) 9.890e-07 0.0009945
bloco (Intercept) 1.970e-01 0.4438035
Number of obs: 1995, groups:
numero, 761; parcela:transecto:bloco, 146; transecto:bloco, 39; bloco, 10
Dispersion parameter for nbinom1 family (): 598
Conditional model:
Estimate Std. Error z value Pr(>|z|)
(Intercept) 8.61419 0.05588 154.16 < 2e-16 ***
ano_medida2017 -0.32062 0.02686 -11.94 < 2e-16 ***
ano_medida2018 -0.47854 0.02501 -19.13 < 2e-16 ***
das_mm 0.08344 0.01869 4.46 8.04e-06 ***
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Zero-inflation model:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.97345 0.22911 -4.249 2.15e-05 ***
ano_medida2017 1.82993 0.18150 10.082 < 2e-16 ***
ano_medida2018 -0.01319 0.17647 -0.075 0.9404
das_mm -1.24866 0.13190 -9.467 < 2e-16 ***
dif_alt_dossel_m -1.30749 0.14970 -8.734 < 2e-16 ***
elevation_m 0.27767 0.17206 1.614 0.1066
hand_m -0.20645 0.15959 -1.294 0.1958
basal_area_m2 -0.71171 0.14941 -4.763 1.90e-06 ***
tree_density 0.33322 0.15064 2.212 0.0270 *
das_mm:dif_alt_dossel_m 0.55660 0.12694 4.385 1.16e-05 ***
elevation_m:hand_m -0.33424 0.13648 -2.449 0.0143 *
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
--
Dr. Alexandre F. Souza
Professor Associado
Departamento de Ecologia/CB
Universidade Federal do Rio Grande do Norte
Campus Universitário - Lagoa Nova
59072-970 - Natal, RN - Brasil
lattes: lattes.cnpq.br/7844758818522706
http://www.esferacientifica.com.br
https://www.youtube.com/user/alexfadigas
http://www.docente.ufrn.br/alexsouza
orcid.org/0000-0001-7468-3631 <http://www.docente.ufrn.br/alexsouza>
[[alternative HTML version deleted]]
_______________________________________________
R-sig-ecology mailing list
R-sig-ecology@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
