Martin-Jung
commented
1 year ago Heya, two possibilities to obtain a 'suitability index':
1)
Simply add pseudo-absences as is common practice. This can be done using add_pseudoabsence() using a method of your choice (random, buffer, etc...). See the help file in pseudoabs_settings.
Example code
# Define pseudo-absence settings
abs <- pseudoabs_settings(background = background, nrpoints = 10000, method = "mcp",inside = FALSE)
# Convert presence-only points in sf format to presence-absence
# Field_occurrence points to a
poa <- my_gbifpoints |> add_pseudoabsence(field_occurrence = "Observed",
template = background,
settings = abs # Settings from above
)
# Train a model
fit <- distribution(background) |>
# Presence-absence data
add_biodiversity_poipa(poa) |>
add_predictors(predictors) |>
engine_glmnet() |>
train()
plot(fit)
write_output(fit, "myprediction.tif")
2) Fit a model using a Poisson Process model. Here the output unit is number of observations per unit area, thus needs to be transformed / normalized if you need a layer in units 0-1. Example code
# Train a model
fit <- distribution(background) |>
# Presence-absence data
add_biodiversity_poipo(my_gbifpoints) |>
add_predictors(predictors) |>
engine_glmnet() |>
train()
plot(fit) # This might have different units
# Get the prediction
pred <- fit$get_data() |>
predictor_transform(option = "norm")I will add soon an option to allow predictor_transform() also on the direct output from train to ease this up a bit as well as different options for normalization. Currently this is done via (x - min) / (max - min)
Cheers, Martin
ManuelSpinola
I really like the package.
I am working with presence-only and oportunistic data, for example, from GBIF.
How can I obtain the output and map from an specific engine in habitat suitability index (between 0 and 1).