webSDM (version 1.1-5)

evaluateModelFit: Evaluates prediction goodness of fit

Description

Evaluate goodness of fit by comparing a true versus a predicted dataset of species distribution. Ypredicted is typically predicted using a prediction method of trophicSDM (in cross-validation if trophicSDM_CV() is used).

Usage

evaluateModelFit(tSDM, Ynew = NULL, Ypredicted = NULL)

Value

A table specifying the goodness of fit metrics for each species. For presence-absence data, the model computes TSS and AUC. For Gaussian data, the R2.

Arguments

tSDM

A trophicSDMfit object obtained with trophicSDM().

Ynew

A sites x species matrix containing the true species occurrences state. If set to NULL (default), it is set to the species distribution data Y on which the model is fitted.

Ypredicted

A sites x species matrix containing the predicted species occurrences state. If set to NULL (default), it is set to the fitted values, i.e. predictions on the dataset used to train the model.

Author

Giovanni Poggiato

References

Grace, J. B., Johnson, D. J., Lefcheck, J. S., and Byrnes, J. E. K.. 2018. Quantifying relative importance: computing standardized effects in models with binary outcomes. Ecosphere 9(6):e02283.

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 = 20,
               family = binomial(link = "logit"), penal = NULL, 
               mode = "prey", method = "stan_glm")
# Evaluate the quality of model predictions on the training
# Predict (fullPost = FALSE) as we used stan_glm to fit the model
# but here we are only intested in the posterior mean
Ypred = predict(m, fullPost = FALSE)
# format predictions to obtain a sites x species dataset whose
# columns are ordered as Ynew
Ypred = do.call(cbind,
                lapply(Ypred, function(x) x$predictions.mean))
                
Ypred = Ypred[,colnames(Y)]
evaluateModelFit(m, Ynew = Y, Ypredicted = Ypred)

# Note that this is equivalent to `evaluateModelFit(m)`
# If we fitted the model using "glm"
m = trophicSDM(Y, X, G, env.formula,
               family = binomial(link = "logit"), penal = NULL, 
               mode = "prey", method = "glm")
Ypred = predict(m, fullPost = FALSE)
# format predictions to obtain a sites x species dataset whose
# columns are ordered as Ynew
Ypred = do.call(cbind, Ypred)
Ypred = Ypred[,colnames(Y)]

evaluateModelFit(m, Ynew = Y, Ypredicted = Ypred)
# Note that this is equivalent to:
# \donttest{
evaluateModelFit(m)
# }