High-Performance Open-Source Archive
snake_case functions: gwpr(),
select_bandwidth(), fit_gwpr(),
diagnose_gwpr().sf first: spatial input must be an
sf object. The legacy sp /
SpatialPolygonsDataFrame interface of the 0.x API is not
supported by the new functions.GWPR(),
bw.GWPR(), GWPR.moran.test(),
GWPR.pFtest(), GWPR.phtest(), and
GWPR.plmtest() remain in the package for backwards
compatibility but will be removed in a future version.gwpr(): one-call pipeline that
validates inputs, optionally searches for the optimal bandwidth, fits
the model, and runs diagnostics.select_bandwidth(): standalone
bandwidth optimisation with three search strategies:
"grid" — exhaustive grid search over a user-defined
range."sgd" — stochastic gradient descent (fast, scales to
large grids)."random" — random sampling over a candidate set.fit_gwpr(): fit a GWPR model with a
known bandwidth. Supports family = "gaussian" (linear) and
family = "binomial" (panel logistic).diagnose_gwpr(): collects results from
Moran’s I ("moran"), local F-test ("f_test"),
Hausman test ("hausman"), and LM test
("lm_test")."bisquare"
(default), "gaussian", "exponential",
"tricube", "boxcar".adaptive = TRUE)
using k-nearest-neighbour approach.workers
argument (requires future and future.apply;
default workers = 1 is always safe).gwpr() warns when
the estimated memory footprint is high before starting long
computations.seed argument
propagates to all stochastic steps.fixest, glmmTMB (for binomial panel
logistic models).future and future.apply moved to
Suggests (only needed for
workers > 1).R (>= 3.5.0) minimum version (was
R (>= 2.10)).
Need mirroring services?
Contact our team at info@vpspulse.com.
Mirror powered by VPSpulse
Infrastructure sponsored by VPSPulse & Secure Payments by ArionPay.