NEWS

---

spatstat.linnet 3.2-3 (2024-11-19)

OVERVIEW

• Shortest path between two points on a network.

NEW FUNCTIONS

• shortestpath Find the shortest path between two specified points on a network, and return it as a line segment pattern.

spatstat.linnet 3.2-2 (2024-09-20)

OVERVIEW

• Tweaks to documentation.
• Internal improvements.

spatstat.linnet 3.2-1 (2024-07-15)

OVERVIEW

• Internal tweaks to satisfy CRAN package checker.

spatstat.linnet 3.2-0

OVERVIEW

• spatstat.linnet now depends on 'spatstat.univar'.
• extract the subset of an image that lies on selected segments of network.
• Minor improvements.
• Internal improvements.

PACKAGE DEPENDENCE

• spatstat.linnet Now depends on the new package 'spatstat.univar'.

SIGNIFICANT USER-VISIBLE CHANGES

• identify.lpp Automatically starts a new plot device and displays 'x' if there is no plot device open.
• thinNetwork X can be a pixel image on a linear network (object of class 'linim').
• plot.linnet, plot.lpp These functions now recognise the argument 'adj.main'.

spatstat.linnet 3.1-5 (2024-03-25)

OVERVIEW

• Slightly accelerated.
• Internal stability improvements.

SIGNIFICANT USER-VISIBLE CHANGES

• spatstat.linnet package Some computations slightly accelerated.

spatstat.linnet 3.1-4 (2024-02-04)

OVERVIEW

• Added a full list of functions to the main help file.
• Intersection of a 2D tessellation with a network.
• Internal improvements.

SIGNIFICANT USER-VISIBLE CHANGES

• intersect.lintess Can compute the intersection between a two-dimensional tessellation and a linear network (yielding a tessellation on the network).

spatstat.linnet 3.1-3 (2023-10-28)

OVERVIEW

• Minor corrections to help files.

spatstat.linnet 3.1-2

OVERVIEW

• Minor improvements to help files.

spatstat.linnet 3.1-1 (2023-05-15)

OVERVIEW

• Minor improvements

SIGNIFICANT USER-VISIBLE CHANGES

• rhohat.lpp, rhohat.lppm New argument 'rule.eps' passed to 'as.mask'.
• as.linim.default, as.linim.linfun New argument 'rule.eps' passed to 'as.mask'.

spatstat.linnet 3.1-0 (2023-04-14)

OVERVIEW

• Inhomogeneous K and pcf using automatic estimation of intensity.
• density.lpp accepts bandwidth selection rules.
• density.lpp accepts sigma=NULL and has a simple default.
• Updated package documentation.

SIGNIFICANT USER-VISIBLE CHANGES

• linearKinhom, linearpcfinhom, linearKEuclidInhom, linearpcfEuclidInhom Argument 'lambda=NULL' is now interpreted to mean that the intensity should be estimated by kernel smoothing. A warning is issued that this is different from the previous behaviour.
• linearKdot.inhom, linearpcfdot.inhom Argument 'lambdaI=NULL' or 'lambdadot=NULL' is now interpreted to mean that the intensity should be estimated by kernel smoothing.
• linearKcross.inhom, linearpcfcross.inhom Argument 'lambdaI=NULL' or 'lambdaJ=NULL' is now interpreted to mean that the intensity should be estimated by kernel smoothing.
• density.lpp Argument 'sigma' can now be a function in the R language, assumed to provide a bandwidth selection rule. This function will be applied to the point pattern x to compute the bandwidth.
• density.lpp Argument 'sigma=NULL' is now accepted. The default value is one-eighth of the length of the shortest side of the bounding box of x.
• densityQuick.lpp Argument 'X' changed to 'x' for consistency.
• linearpcfinhom New arguments 'adjust.sigma', 'bw' and 'adjust.bw'.
• linearpcfdot.inhom, linearpcfcross.inhom New arguments 'adjust.sigma', 'bw' and 'adjust.bw'.
• linearpcfEuclidInhom New arguments 'adjust.sigma', 'bw' and 'adjust.bw'.

spatstat.linnet 3.0-6 (2023-02-22)

OVERVIEW

• Internal improvements and bug fixes.

spatstat.linnet 3.0-5

OVERVIEW

• Improvements and bug fixes in 'lppm' class.
• Minor extension to random generators.

SIGNIFICANT USER-VISIBLE CHANGES

• runiflpp, rpoislpp The simulation parameters can be determined from an example point pattern, given as the argument 'ex'.
• print.lppm, summary.lppm The name of the original point pattern dataset (to which the model was fitted) is now printed.
• print.lppm, summary.lppm Improved output.
• update.lppm Internal improvements.

spatstat.linnet 3.0-4 (2023-01-27)

OVERVIEW

• quantile function on a network.
• Minor enhancements and bug fixes.

NEW FUNCTIONS

• quantilefun.linim Quantile function, returned as a function. Method for 'quantilefun' for images on a network.

BUG FIXES

• relrisk.lpp Crashed if allow.infinite=FALSE. Fixed.

spatstat.linnet 3.0-3 (2022-11-15)

OVERVIEW

• We thank Brian Ripley for contributions.
• Internal bug fix.

BUG FIXES

• pairdist.lpp Results could have been completely incorrect, due to an internal bug, if the linear network data was in the non-sparse representation.

spatstat.linnet 3.0-2 (2022-11-09)

OVERVIEW

• Internal tweaks to satisfy package checker.

spatstat.linnet 3.0-1

OVERVIEW

• Internal improvements.

spatstat.linnet 3.0-0

OVERVIEW

• We thank Greg McSwiggan, Fernando Milesi, Mehdi Moradi, Gopalan Nair and James Pope for contributions.
• spatstat.linnet now depends on the new packages 'spatstat.explore' and 'spatstat.model' rather than the old 'spatstat.core'.
• Kernel smoothing on a linear network.
• K function and pair correlation function based on Euclidean distance.
• Inhomogeneous J function on a network.
• Terminal vertices of a network.
• Points along a network can be drawn as cross-ticks.
• Marks attached to vertices and segments of a network.
• Important change to K function and pair correlation function when 'ratio=TRUE'.
• U-shaped curves in 'rhohat'.
• Faster computation of the length of the disc in the shortest path metric.
• Minor improvements and bug fixes.
• Improvements to internal code.
• Minor internal changes to package information.

NEW FUNCTIONS

• lineardisclength Compute the length of the disc in the shortest-path metric.
• marks.linnet, marks<-.linnet Marks attached to the vertices and segments of a linear network.
• Smooth.lpp Kernel smoothing (Nadaraya-Watson) on a linear network.
• distmap.lpp Distance map for a point pattern on a network.
• linearKEuclid, linearpcfEuclid, linearKEuclidInhom, linearpcfEuclidInhom K function and pair correlation function based on Euclidean distances.
• linearJinhom Inhomogeneous J function on a linear network.
• terminalvertices Extract the terminal vertices of a linear network.
• bw.relrisk.lpp This function replaces 'bw.relrisklpp' and is a method for the generic 'bw.relrisk'.
• rThomaslpp Simulate Thomas cluster process on a network.

SIGNIFICANT USER-VISIBLE CHANGES

• Package dependence 'spatstat.linnet' now depends on the new packages 'spatstat.explore' and 'spatstat.model' rather than the old 'spatstat.core'.
• plot.lpp If 'shape="crossticks"', the points will be drawn as short line segments perpendicular to the network.
• linearK, linearpcf, linearKdot, linearKcross, linearpcfcross, linearpcfdot When 'ratio=TRUE', the denominator is now equal to the number of ordered pairs of data points considered. This does not affect the summary function, but it changes the calculation of pooled estimates.
• densityQuick.lpp If 'sigma' is a function for selecting a bandwidth, additional arguments '...' of densityQuick.lpp will be passed to 'sigma', provided they match the name of a formal argument of 'sigma'.
• bw.relrisklpp This function is now deprecated, and is replaced by 'bw.relrisk.lpp', a method for the generic 'bw.relrisk'.
• bw.relrisk.lpp When 'method="likelihood"', the cross-validation criterion is now defined as the *negative* likelihood. This is consistent with 'bw.relrisk.ppp', and ensures that the optimum bandwidth is always found by minimising the cross-validation criterion.
• integral.linim, integral.linfun New argument 'weight' specifies a weight function for the integration.
• rhohat.lpp, rhohat.lppm New options 'smoother="mountain"' and 'smoother="valley"' for estimating a unimodal function (U-shaped curve).
• rhohat.lpp, rhohat.lppm If the covariate is a 'distfun' on a network, the unit of length will be saved, and will be displayed on the plot.
• rhohat.lpp, rhohat.lppm New arguments 'jitter', 'jitterfactor', 'interpolate' allow greater control over the calculation.
• plot.linnet Optionally displays the marks of the segments or marks of the vertices. Changed argument list.
• print.linnet, summary.linnet These functions print information about marks of the segments or vertices.
• predict.lppm Can now compute standard errors.
• predict.lppm New formal arguments 'covariates' and 'se'.
• rhohat.lpp, rhohat.lppm New argument 'do.CI' specifies whether to calculate confidence bands.

BUG FIXES

• density.lpp If 'weights' were given, the results were completely incorrect when leaveoneout=TRUE (the default) and at="points".
• spatstat.model::effectfun Results were sometimes incorrect for 'lppm' objects. [Spotted by Fernando A. Milesi] Fixed.
• predict.lppm Ignored argument 'covariates'. Fixed.
• rhohat.lpp, rhohat.lppm The argument 'subset' was not handled correctly in the internal data. The estimated function 'rho' was correct but the results of 'predict.rhohat' may have been incorrect and were computed only on the 'subset'. Fixed.
• plot.lpp Did not display the main title in some cases. Fixed.
• "[.lpp", "[.linnet" Ignored marks attached to the network segments. Fixed.
• "[.lpp", "[.linnet" Crashed if the network vertices had marks, when 'snip=TRUE'. Fixed.
• densityQuick.lpp Crashed sometimes with a message about unequal lengths of vectors. Fixed.

spatstat.linnet 2.3-2 (2022-02-16)

OVERVIEW

• 'spatstat.linnet' now depends on 'spatstat.random'.

SIGNIFICANT USER-VISIBLE CHANGES

• package dependence 'spatstat.linnet' now depends on the new package 'spatstat.random'.

spatstat.linnet 2.3-1 (2021-12-11)

OVERVIEW

• More control over resolution of 'linim' objects.
• Improved documentation.
• Minor improvements and bug fixes.

SIGNIFICANT USER-VISIBLE CHANGES

• as.linim.default, as.linim.linfun New argument 'nd'
• integral.linfun New argument 'nd'
• rjitterlpp, rjitter.lpp The function 'rjitterlpp' has been renamed 'rjitter.lpp' and is now a method for the generic 'rjitter'.
• rjitterlpp This function still exists, but is now deprecated in favour of 'rjitter.lpp'.

BUG FIXES

• rjitterlpp The argument 'radius' was interpreted as a fraction of segment length rather than an absolute distance. Fixed.

spatstat.linnet 2.3-0 (2021-07-17)

OVERVIEW

• We thank Suman Rakshit, Greg McSwiggan and Marc Schneble for contributions.
• Sufficient Dimension Reduction on a linear network.
• Perspective-view plots of linfun and linim objects.
• Bug fix to Math.linim
• Minor additions.

NEW FUNCTIONS

• sdr.lpp Sufficient Dimension Reduction on a linear network. [Contributed by Suman Rakshit.]
• persp.linfun, persp.linim Perspective-view plots of functions on a linear network (class linim) and images on a linear network (class linfun). [Written by Adrian Baddeley and Greg McSwiggan.]

SIGNIFICANT USER-VISIBLE CHANGES

• response.lppm Method for 'response' for fitted point process models on a network.

BUG FIXES

• Math.linim If 'Z' was a pixel image on a network (class 'linim') and 'v' was a single number, then 'Z/v' was a pixel image (class 'im') instead of a pixel image on a network (class 'linim'). [Spotted by Marc Schneble.] Fixed.
• bw.voronoi Printed output stated that the cross-validation criterion was minimised instead of maximised.

spatstat.linnet 2.2-1 (2021-06-22)

OVERVIEW

• Minor changes to satisfy CRAN.

spatstat.linnet 2.2-0

OVERVIEW

• Extensions to rhohat.lpp and rhohat.lppm.
• Internal bug fixes.

SIGNIFICANT USER-VISIBLE CHANGES

• rhohat.lpp, rhohat.lppm New option (smoother='piecewise') computes a piecewise-constant estimate of rho(z).
• rhohat.lpp, rhohat.lppm The result now includes the 'average' intensity rho.

spatstat.linnet 2.1-1 (2021-03-28)

OVERVIEW

• Tweak to satisfy the package checker.

spatstat.linnet 2.1-0

OVERVIEW

• We thank Andrea Gilardi for contributions.
• Bug fix in density.lpp
• Improvement to densityEqualSplit
• New function for jittering point patterns on a network.
• Function renamed.

NEW FUNCTIONS

• rjitterlpp Apply random displacements to the points on a linear network.
• densityHeat.lpp Diffusion kernel estimation of intensity on a linear network. (formerly known as 'densityHeatlpp' and 'densityHeat', this is now a method for the new generic 'densityHeat')

SIGNIFICANT USER-VISIBLE CHANGES

• densityEqualSplit New arguments 'at' and 'leaveoneout' for consistency with other functions.
• densityHeat The function formerly known as 'densityHeat' or 'densityHeatlpp' is now renamed 'densityHeat.lpp' and is a method for the generic 'densityHeat'.
• density.lpp Accelerated when the pattern contains duplicated points.

BUG FIXES

• density.lpp The result had the wrong length if 'x' contained duplicated points when 'weights' were given and 'at="points"'. [Spotted by Andrea Gilardi] Fixed.

spatstat.linnet 2.0-0 (2021-03-18)

OVERVIEW

• We thank Andrea Gilardi for contributions.
• Function renamed.
• Bug fix in density.lpp

SIGNIFICANT USER-VISIBLE CHANGES

• densityHeat 'densityHeat' has been temporarily renamed 'densityHeatlpp', to prevent errors in the package checking process.
• [.linim Accelerated.

BUG FIXES

• deviance.lppm, pseudoR2.lppm Results were completely incorrect, due to a coding error. Fixed.
• density.lpp Crashed if 'weights' were given and 'x' contained duplicated points. [Spotted by Andrea Gilardi] Fixed.
• Lcross.inhom, Kcross.inhom, Kmulti.inhom The option 'correction="none"' was accepted but ignored. [Spotted by Corey Anderson.] Fixed.
• simulate.rhohat Crashed when applied to rhohat objects computed from data on a linear network. Fixed.

spatstat.linnet 1.65-9

OVERVIEW

• Minor internal changes.

spatstat.linnet 1.65-8

OVERVIEW

• Function renamed.

SIGNIFICANT USER-VISIBLE CHANGES

• densityHeat 'densityHeat' has been renamed 'densityHeat.lpp'.

spatstat.linnet 1.65-7

OVERVIEW

• Tweaks to satisfy CRAN.

spatstat.linnet 1.65-1

OVERVIEW

• Added NEWS file.

spatstat.linnet 1.65-0

OVERVIEW

• Package initialised at version 1.65-0

SIGNIFICANT USER-VISIBLE CHANGES

• spatstat.linnet The package 'spatstat.linnet' has been created from a subset of the code in the original 'spatstat' package version 1.65-0. It contains the functionality for statistical analysis of spatial data on a linear network. For an overview, see help("spatstat.linnet-package")
• Execution The 'spatstat.linnet' package is slightly faster than the corresponding code in the 'spatstat' package, because the procedure for calling internal C functions has been streamlined.