NEWS.md

In MarcSchneble/geonet: Intensity Estimation on Geometric Networks with Penalized Splines

geonet version 0.7.3 (CRAN and GitHub release)

• selective exporting of functions
• checking for inputs

geonet version 0.6.0 (CRAN and GitHub release)

• revised plotting functions (caution: new arguments defining line widths, etc.)
• scale of the dist2V variable is now supported
• added argument "other arguments" to intensity_pspline for compatibility with older versions
• revised summary for gnppfit objects

geonet version 0.5.0 (Github release)

• new function network_ISE for the computation of integrated squared errors
• new function network_integral for the computation of an integral of the intensity
• geonet supplies wrappers for density.lpp, i.e. the intensity on a geometric network can be estimated by employing kernel based methods
• fixed bug which occurs if a smooth covariate is specified in formula after a linear covariate

geonet version 0.4.0 (Github release)

• summaries and related print methods now also print information on the linear network representation
• default penalty is now of order r = 2
• the global knot distance delta can now be supplied as a quantile of the curve lengths of the network and the global bin width h as a fraction of the global knot distance
• fixed bug in as_gnpp.lpp when linear point pattern is not marked
• fixed bug in runif_gn
• added function rgnpp which allows to simulate from a fitted intensity

geonet version 0.3.0 (Github release)

• algorithm options can now be supplied to intensity_pspline via the "control" argument
• as_gn now also takes the "units" attribute from a linnet object
• added missing summary and print methods
• fixed bug in "internal" occurring when a linear covariates has length 1
• internal covariates are recognized automatically and do not need to be assigned in the formula

geonet version 0.2.0 (GitHub release)

• added internal covariate information to the montgomery network
• fixed bug in network_penalty occurring when r = 1
• model summary has been changed
• fit_poisson_model is only doing on Fisher scoring iteration within the inner loop from the second iteration
• effective degrees of freedom is computed for smooth terms
• allow for general internal covariates added to the lins attribute of the network
• internal covariate "dist2V" can be added to the linear predictor
• stopping criterion now depends on relative difference of theta and not of rho
• verbose argument added to intensity_pspline which allows to track the progress of the fitting algorithm

geonet version 0.1.1 (CRAN and GitHub release)

• Added reference in the description file.
• Added missing return values in the documentation.
• Removed set.seed() call.

geonet version 0.1.0 (no release)

• Initial submission

MarcSchneble/geonet documentation built on Jan. 28, 2023, 7:49 p.m.