webSDM (version 1.1-5)

predict.trophicSDMfit: Computes predicted values from the fitted trophicSDMfit model

Description

Computes predicted values from the fitted trophicSDMfit model at environmental conditions specified by Xnew. Once predictions have been obtained, their quality can eventually be evaluated with evaluateModelFit().

Usage

# S3 method for trophicSDMfit
predict(
  object,
  Xnew = NULL,
  prob.cov = FALSE,
  pred_samples = NULL,
  run.parallel = FALSE,
  verbose = FALSE,
  fullPost = TRUE,
  filter.table = NULL,
  ...
)

Value

A list containing for each species the predicted value at each sites. If method = "stan_glm", then each element of the list is a sites x pred_samples matrix containing the posterior predictive distribution of the species at each sites.

Arguments

object

A trophicSDMfit object obtained with trophicSDM()

Xnew

a matrix specifying the environmental covariates for the predictions to be made. If NULL (default), predictions are done on the training dataset (e.g. by setting Xnew = tSDM$data$X).

prob.cov

Parameter to predict with trophicSDM with presence-absence data. Whether to use predicted probability of presence (prob.cov = T) or the transformed presence-absences (default, prov.cov = F) to predict species distribution.

pred_samples

Number of samples to draw from species posterior predictive distribution when method = "stan_glm". If NULL, set by the default to the number of iterations/10.

run.parallel

Whether to use parallelise code when possible. Can speed up computation time.

verbose

Whether to print advances of the algorithm

fullPost

Optional parameter for stan_glm only. Whether to give back the full posterior predictive distribution (default, fullPost = TRUE) or just the posterior mean, and 2.5% and 97.5% quantiles,

filter.table

Optional, default to NULL, should be provided only if the users wants to filter some species predictions. A sites x species matrix of zeros and ones.

...

additional arguments

Author

Giovanni Poggiato and Jérémy Andréoletti

Examples

data(Y, X, G)
# define abiotic part of the model
env.formula = "~ X_1 + X_2"
# Run the model with bottom-up control using stan_glm as fitting method and no penalisation
# (set iter = 1000 to obtain reliable results)
m = trophicSDM(Y, X, G, env.formula, iter = 50,
               family = binomial(link = "logit"), penal = NULL, 
               mode = "prey", method = "stan_glm")
# We can now evaluate species probabilities of presence for the environmental conditions c(0.5, 0.5)
predict(m, Xnew = data.frame(X_1 = 0.5, X_2 = 0.5))
# Obtain 50 draws from the posterior predictive distribution of species (pred_samples = 10)
# using predicted presence-absences of species to predict their predators (prob.cov = TRUE)
# Since we don't specify Xnew, the function sets Xnew = X by default
Ypred = predict(m, fullPost = TRUE, pred_samples = 10, prob.cov = FALSE)
# We can ask the function to only give back posterior mean and 95% credible intervals with
# fullPost = F
# \donttest{
Ypred = predict(m, fullPost = TRUE, pred_samples = 30, prob.cov = FALSE)
# }
# If we fit the model using in a frequentist  way (e.g. glm)
m = trophicSDM(Y, X, G, env.formula, 
               family = binomial(link = "logit"), penal = NULL, 
               mode = "prey", method = "glm")
# We are obliged to set pred_samples = 1 
# (this is done by default if pred_samples is not provided)
# In the frequentist case, fullPost is useless.
 Ypred = predict(m, pred_samples = 1, prob.cov = FALSE)