h3sdm_pa() now transforms presence records to the CRS
of the H3 grid before joining, fixing an error when the grid is in a
projected CRS.h3sdm_aoa() estimates the Dissimilarity Index (DI) and
the Area of Applicability (AOA) for spatial prediction models, based on
Meyer & Pebesma (2021).h3sdm_get_grid() now preserves the CRS of the input
sf_object. Previously, the function always returned the
grid in WGS84 (EPSG:4326) regardless of the input CRS. Now, if the AOI
is in a projected CRS, the output grid will be reprojected to match it.
The internal H3 computation still uses WGS84 as required by the H3
system.h3sdm_predict() internal comments translated to English
and @seealso updated to include
h3sdm_aoa().h3sdm_fit_model() now automatically detects model
mode (classification or regression), enabling count-based models
(Poisson, Negative Binomial) with appropriate metrics (RMSE, R², MAE)
without requiring manual configuration.
h3sdm_fit_model() and h3sdm_predict()
now automatically detect model mode (classification or regression),
enabling count-based models (Poisson, Negative Binomial) without manual
configuration. Full backward compatibility maintained.
h3sdm_get_records() now supports
"biodatacr" as an optional provider, querying occurrence
records from BiodataCR (Costa Rica) via the rbiodatacr
package. h3sdm_pa() inherits this support automatically
through its providers argument. rbiodatacr is
listed as a suggested dependency.
h3sdm_pa_from_records(): generates a
presence/pseudo-absence dataset from user-provided records. Accepts a
data.frame or sf object with coordinates in
WGS84 (EPSG:4326). Supports optional filtering by a
geospatialKosher column to remove records with questionable
spatial quality.
h3sdm_count_from_records(): generates a hexagonal
grid with count-based response variables (species richness, total
detections, or individual abundance) from user-provided records. Accepts
a data.frame or sf object. Supports optional
filtering by presence column, confidence threshold, and date
range.
h3sdm_recipe(): added response_col
parameter (default "presence") to support count-based
response variables. Use response_col = "count" when working
with data generated by h3sdm_count_from_records().
h3sdm_recipe_gam(): added response_col
parameter (default "presence") with the same behavior as
h3sdm_recipe(). Also added documentation examples for both
presence/absence and count-based models.
h3sdm_workflow_gam(): updated documentation to
clarify the use of set_mode("classification") for
presence/absence models and set_mode("regression") with
family = poisson() for count-based models.
h3sdm_workflow(): updated documentation to clarify
model mode selection for presence/absence and count-based
models.
h3sdm_workflows(): updated documentation to clarify
model mode selection for presence/absence and count-based
models.