Package index • spatialkit

Overview

spatialkit spatialkit-package: spatialkit: Spatial Tessellation, Modeling, and Cross-Validation Toolkit

Prepare the data

Coordinate reference systems, geometry coercion, and the clip target.

ensure_projected(): Ensure an object has a projected CRS (with sensible defaults)
harmonize_crs(): Harmonize CRS between two spatial objects
coerce_to_points(): Coerce arbitrary geometries to representative points
clip_target_for(): Build a polygonal clip target from points and/or a boundary
prep_model_data(): Prepare and sanitize an sf dataset for spatial modeling

Measure the spatial structure

The distance over which observations stay correlated. It sizes both the cells and the cross-validation blocks.

estimate_sac_range(): Estimate the spatial autocorrelation range from data
sac_nugget(): The nugget of an estimated autocorrelation range
print(<sac_range>): Print a spatial autocorrelation range
plot(<sac_range>): Plot an estimated spatial autocorrelation range

Choose a resolution

How many cells, and how sure that number is.

determine_optimal_levels(): Determine an optimal number of spatial levels via an elbow heuristic
resolution_profile(): Score every candidate number of cells on several criteria at once
select_resolution(): Read a level, and the region over which it is not distinguishable, off a profile
summary(<resolution_profile>): Every criterion's pick, side by side
print(<resolution_profile>): Print a resolution profile
plot(<resolution_profile>): Plot a resolution profile

Tessellate

Build the regions, with reproducible cell identifiers.

build_tessellation(): Build a tessellation (Voronoi, Delaunay triangles, hex grid, or square grid)
get_voronoi_seeds(): Generate seed points for Voronoi tessellation
voronoi_seeds_kmeans(): K-means seed generation from point coordinates
voronoi_seeds_random(): Random seed generation within a polygonal boundary
create_voronoi_polygons(): Create Voronoi polygons from points with CRS handling and optional clipping
create_grid_polygons(): Create square or hexagonal grid polygons over a boundary
create_grid_polygons_cached(): Create and cache grid polygons over a boundary
ensure_stable_poly_id(): Create deterministic, stable polygon IDs based on spatial sort keys
plot_tessellation_map(): Plot a tessellation map with optional boundary, seeds, and features

Assign and aggregate

One row per cell, with a count and a standard error for every aggregate, the record of which rows were dropped or tied on the way, and whether a block-kriging estimate would have beaten the plain cell mean.

assign_features_to_polygons(): Assign features to polygons and attach a polygon ID
summarize_by_cell(): Summarize features by polygon/cell ID
kriging_adequacy(): Block-kriging adequacy diagnostics for a set of cells
`[`(<spatialkit_rows>): Subset a layer that carries a row record

Fold

Spatial cross-validation folds, how large the blocks should be, and how far the held-out points ended up from the training data.

make_folds(): Create spatial cross-validation folds
plot_folds(): Map a cross-validation fold scheme
fold_separation(): How far the held-out points actually sit from the training data
cv_block_size_sweep(): Cross-validate at a ladder of block sizes
plot(<block_size_sweep>): Plot cross-validation error against block size

Fit

Three backends behind one spatial_fit class, and the constructor for a backend of your own.

new_spatial_fit(): Build a spatial_fit S3 object
fit_rf_model(): Fit a random forest via ranger
fit_gwr_model(): Fit a Geographically Weighted Regression (GWR) via GWmodel
fit_bayesian_spatial_model(): Fit a Bayesian spatial regression with a 2D Gaussian Process (via brms)
gp_lengthscale_bounds(): Heuristic length-scale bounds for a squared-exponential GP

Choose the predictors

Forward selection scored on spatial folds for any backend, and an AICc search over predictor sets for GWR.

select_features_forward(): Greedy forward feature selection with spatially blocked inner folds
plot(<feature_selection>): Plot the path of a forward feature selection
gwr_model_selection(): Forward model selection for geographically weighted regression
print(<gwr_model_selection>): Print a GWR model selection result
plot(<gwr_model_selection>): Plot a GWR model selection

Methods on a fit

print(<spatial_fit>): Print a fitted spatial model
summary(<spatial_fit>): Summarise a fitted spatial model
plot(<spatial_fit>): Plot a fitted spatial model
predict(<rf_fit>): Predict from a random forest fit
predict(<gwr_fit>): Predict from a GWR spatial model
predict(<bayesian_fit>): Predict from a Bayesian spatial GP model
fitted(<rf_fit>): Out-of-bag predictions from a random forest fit
fitted(<gwr_fit>): In-sample fitted values from a GWR fit
fitted(<bayesian_fit>): In-sample fitted values from a Bayesian spatial GP fit
residuals(<rf_fit>): Out-of-bag residuals from a random forest fit
residuals(<gwr_fit>): In-sample residuals from a GWR fit
residuals(<bayesian_fit>): In-sample residuals from a Bayesian spatial GP fit
coef(<rf_fit>): Coefficients are undefined for a random forest
coef(<gwr_fit>): Extract GWR local coefficients
coef(<bayesian_fit>): Extract Bayesian model fixed-effect summaries
print(<rf_fit>): Print a random forest fit

Read a fit in sample

Metrics on the training data (out-of-bag for a forest), and the residual checks that say whether structure was left behind. In-sample numbers are not comparable across backends; compare_models_cv() is.

model_metrics(): Compute goodness-of-fit metrics for a spatial model
evaluate_insample(): Compute in-sample (or out-of-sample) metrics for fitted spatial models
compare_models(): Side-by-side comparison of fitted spatial models
residual_morans_i(): Compute Moran's I on the residuals of a fitted spatial model
print(<morans_i>): Print a residual Moran's I result

Validate

Score on held-out blocks, and compare backends on the same folds.

cv_spatial(): Model-agnostic spatial cross-validation
cv_rf(): Cross-validate a random forest with spatial folds
cv_gwr(): K-fold cross-validation for GWR
cv_bayes(): K-fold cross-validation for the Bayesian spatial model
compare_models_cv(): Cross-validated comparison of spatial models
plot_cv_metrics(): Plot one cross-validation metric fold by fold
plot_calibration(): Plot the interval calibration of a Bayesian cross-validation

Predict, and check where the prediction applies

predict_surface(): Predict a fitted spatial model onto a regular grid
area_of_applicability(): Area of applicability of a spatial prediction model
print(<aoa>): Print an area-of-applicability result
plot(<aoa>): Plot the dissimilarity distribution behind an area of applicability

Package options and caches

The console log, and the two in-session caches. Both caches are keyed on the inputs, so they only need clearing when memory is the constraint.

spatialkit_quiet(): Quieten (or restore) spatialkit's console log
clear_grid_cache(): Clear the in-session grid cache
clear_fitted_cache(): Clear cached fitted values for a Bayesian spatial model