[R-sig-ME] Predicting values from MCMCglmm model with statistical weight in mev argument
Jarrod Hadfield
j@h@d||e|d @end|ng |rom ed@@c@uk
Tue Feb 18 12:05:17 CET 2020
Hi Kamal,
Can you post your sessionInfo()?
As a work around, use this model
model <- MCMCglmm(lD ~ tactic*period*seasonality+complique_KF+lbody+lintdur+lnb.loc+lduration
, random = ~sp_phylo+species2+phylo_pop+phylo_popY+phylo_pop_id + idh(SE):units,
, family = "gaussian"
, ginverse = list(sp_phylo = inv.phylo$Ainv) # include a custom matrix for argument phylo
, prior = prior1
, data = Data
, nitt = 22e+04
, burnin = 20000
, thin = 100
, pr=TRUE)
BUT make sure to fix the prior variance associated with the final random effect term (idh(SE):units) to one. Its identical to the model you've fitted, but the predict function should work.
Cheers,
Jarrod
On 15/02/2020 22:57, Kamal Atmeh wrote:
model <- MCMCglmm(lD ~ tactic*period*seasonality+complique_KF+lbody+lintdur+lnb.loc+lduration
, random = ~sp_phylo+species2+phylo_pop+phylo_popY+phylo_pop_id
, family = "gaussian"
, mev = SE^2 # error variance associated to each data point
, ginverse = list(sp_phylo = inv.phylo$Ainv) # include a custom matrix for argument phylo
, prior = prior1
, data = Data
, nitt = 22e+04
, burnin = 20000
, thin = 100
, pr=TRUE)
The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.
[[alternative HTML version deleted]]
More information about the R-sig-mixed-models
mailing list