CVS commit: pkgsrc/geography/R-spatstat.model

["Makoto Fujiwara" <mef%netbsd.org@localhost>]

Module Name:    pkgsrc
Committed By:   mef
Date:           Fri Jun 12 13:59:24 UTC 2026

Modified Files:
        pkgsrc/geography/R-spatstat.model: Makefile distinfo

Log Message:
(geography/R-spatstat.model) Updated 3.3.4  to 3.7.1

        CHANGES IN spatstat.model VERSION 3.7-1

OVERVIEW

    o We thank Martin Emons for contributions.

    o More methods for 'intensity' and 'repul'.

    o Improved handling of boundary pixels in model predictions.

    o Bug fixes and minor improvements.

NEW FUNCTIONS

    o repul.detpointprocfamily
    Repulsion index for determinantal point process models.
    A new method for the generic 'repul'.

    o intensity.kppm
    Method for 'intensity' for fitted cluster processes and Cox processes.

    o intensity.mppm
    Method for 'intensity' for models fitted to multiple point patterns.

    o repul.kppm, repul.ppm, repul.slrm, repul.mppm, repul.rppm, repul.clusterprocess
    New methods for the generic 'repul' calculate the repulsion index
    for Poisson, Gibbs, Cox and Neyman-Scott point process models.

SIGNIFICANT USER-VISIBLE CHANGES

    o predict.ppm, predict.kppm, predict.dppm, predict.rppm, predict.mppm
    These methods now recognise argument 'rule.pix' passed to owin2mask.
    When the original point pattern dataset was observed in a polygonal window,
    setting rule.pix="cover" ensures that pixels which partially intersect
    the polygon will be included in the resulting pixel image.

    o relrisk.ppm
    Now recognises argument 'rule.pix' passed to owin2mask.
    When the original point pattern dataset was observed in a polygonal window,
    setting rule.pix="cover" ensures that pixels which partially intersect
    the polygon will be included in the resulting pixel image.

    o Kcom, Kres
    Computation accelerated when using Ripley isotropic correction.

    o print.kppm
    The index of clustering strength is now printed when 'x' is
    a log-Gaussian Cox process (as well as when 'x' is a Neyman-Scott
    cluster process).

BUG FIXES

   o vcov.ppm
   There was a loophole in the code which *could* have caused
   the result of vcov.ppm to be incorrect for pairwise interaction models
   in a very rare case. No examples have been found.
   Fixed.

   o mppm
   Generated an error about a subscript out of bounds,
   if the package 'formula.tools' was loaded.
   [Spotted by Martin Emons.]
   Fixed.

   o pcfmodel.ppm
   Crashed for pairwise interaction models with a hard core component.
   Fixed.

        CHANGES IN spatstat.model VERSION 3.7-0

OVERVIEW

    o New generic 'Lmodel'.

    o Methods for class 'clustermodel'

    o 'envelope' method for 'dppm'

    o More options for 'simulate' argument in 'envelope'

    o Specify the 'theoretical' value of the summary function in 'envelope'

NEW FUNCTIONS

    o envelope.dppm
    Method for 'envelope' for fitted models of class 'dppm'

    o Lmodel, Lmodel.kppm, Lmodel.dppm, Lmodel.ppm, Lmodel.detpointprocfamily
    Given a point process model, return a function in the R language
    which computes the theoretical L-function of the model.

    o Kmodel.clustermodel, pcfmodel.clustermodel, Lmodel.clustermodel
    Calculate the theoretical K-function, L-function or pair correlation
    function for a model of class 'clustermodel'.

SIGNIFICANT USER-VISIBLE CHANGES

    o psib, persist, clusterstrength, varcount
    These functions now accept an object of class 'clustermodel'.

    o envelope.ppm, envelope.kppm, envelope.slrm
    Argument 'simulate' can be a point process model of class
    'ppm', 'kppm', 'dppm', 'slrm', 'clusterprocess' or 'detpointprocfamily'.

    o envelope.ppm, envelope.kppm, envelope.slrm
    New argument 'theoryfun' can be used to provide the theoretical value
    of the summary function.

        CHANGES IN spatstat.model VERSION 3.6-1

OVERVIEW

    o Minor changes to satisfy CRAN.

        CHANGES IN spatstat.model VERSION 3.6-0

OVERVIEW

    o We thank Gabriela Calana Somoza, Bethany Macdonald
    and Rasmus Waagepetersen for contributions.

    o Waagepetersen's composite likelihood for cluster processes.

    o Convert a recursively partitioned point process model to a tessellation.

    o Improvements to 'profilepl' and 'effectfun'.

    o Diagnostic functions have argument 'rmax'.

    o Cluster strength index and spatial persistence index.

    o Bug fixes and minor improvements.

NEW FUNCTIONS

    o as.tess.rppm
    Convert a recursively partitioned point process model
    to a spatial tessellation.

    o clusterstrength
    Cluster strength index for a cluster process or Cox process.

    o formula.rppm
    Extract the model formula for a recursively partitioned point process model.

    o persist
    Spatial persistence index for a cluster process.

SIGNIFICANT USER-VISIBLE CHANGES

    o kppm
    New option: method="waag" fits the model by maximising
    Waagepetersen's (2007) second order composite likelihood
    [Suggested by Bethany Macdonald]

    o effectfun
    Improved defaults for a recursive partition model (class 'rppm').
    The user only needs to provide values for those covariates which
    are used in the decision tree. If the entire tree involves only one
    variable, then 'covname' is not needed.

    o effectfun
    If a spatial covariate is constant, its value does not need to be
    provided in the call to effectfun().

    o as.fv.kppm
    Improved arguments for empirical estimation of pair correlation function
    when the model was fitted with a non-default value of 'method'.

    o Gcom, Kcom, psstA, psstG, psst
    These diagnostic functions have a new formal argument 'rmax'.

    o Gres, Kres
    These diagnostic functions accept argument 'rmax'.

    o lurking.ppm, lurking.ppp, lurking.slrm
    Argument 'covariate' can be a function(x,y)
    or one of the strings "x" or "y".

    o lurking.ppm, lurking.ppp, lurking.slrm, plot.lurk
    When the covariate is a cartesian coordinate or a distance function,
    the unit of length is stated on the horizontal axis label of the plot.

    o Penttinen
    Computation of interaction terms has been accelerated.

    o profilepl
    It is now possible to fit a model with a formula like ~ pmax(x, thresh)
    where 'thresh' is an irregular parameter.

    o Strauss
    Computation of interaction terms has been slightly accelerated.

BUG FIXES

   o effectfun
   Crashed if the model involved a covariate function
   with extra arguments passed through 'covfunargs'.
   Fixed.

   o lurking
   Crashed in rare cases when 'envelope=TRUE' with message 'Xsim not found'.
   [Spotted by Gabriela Calana Somoza.]
   Fixed.

   o plot.leverage.ppm
   Return value was NULL.
   Fixed.

   o update.ppm
   update(object, X) crashed sometimes if X was a point pattern
   whose observation window was not identical to the window of the
   original point pattern dataset.
   Fixed.

        CHANGES IN spatstat.model VERSION 3.5-0

OVERVIEW

    o Models for replicated point patterns on linear networks.

    o ROC curves for determinantal point process models.

    o Partial residuals for cluster process and Cox process models.

    o Extended facilities for spatial logistic regression.

    o Bug fixes and internal improvements.

NEW FUNCTIONS

    o diagnose
    New generic function for model diagnostics.

    o parres
    New generic function for partial residuals.

    o parres.ppm
    The original function 'parres' has now become the method 'parres.ppm'.

    o parres.kppm
    Partial residual plot for fitted cluster process or Cox process models.

    o roc.dppm
    Compute ROC curve for a determinantal point process model.

SIGNIFICANT USER-VISIBLE CHANGES

    o diagnose.ppm
    This is now a method for the generic function 'diagnose'.

    o fitted.slrm
    Argument 'leaveoneout' is no longer ignored when 'dataonly=FALSE';
    it now supports 'leave-one-pixel-out' prediction.

    o mppm
    Now allows the response variable to be a point pattern on a linear network.
    This is experimental code.

    o predict.slrm
    New argument 'leaveoneout' supports leave-one-pixel-out calculation
    of predicted values, and new argument 'fast' determines the algorithm used.

    o print.slrm
    Now prints information about the pixel raster.

    o residualMeasure
    More options for argument 'lambda'.

    o roc.slrm
    Changed interpretation of argument 'leaveoneout=TRUE'.
    This now performs 'leave-one-pixel-out' calculation
    (previously it performed 'leave-one-point-out' calculation).

    o slrm
    New argument 'ruleAtPoints' controls discretisation rule at data points.

    o slrm
    Arguments '...' are now passed to 'owin2mask' to determine the
    polygon-to-raster discretisation rule for a covariate of class 'owin'.

    o summary.slrm
    Now prints information about the pixel raster.

BUG FIXES

    o fitted.slrm
    Calculation was incorrect when leaveoneout=TRUE and type != "intensity".
    Fixed.

        CHANGES IN spatstat.model VERSION 3.4-2

OVERVIEW

    o Internal changes to satisfy CRAN.

        CHANGES IN spatstat.model VERSION 3.4-1

OVERVIEW

    o Improvements to intensity.ppm for Geyer models.

    o Simulation of a fitted model in a different window is better supported.

    o Minor improvements.

    o We thank Andrea Gilardi for contributions.

SIGNIFICANT USER-VISIBLE CHANGES

    o intensity.ppm
    For Geyer models, the computation is faster,
    and supports more values of the saturation parameter 'sat',
    namely sat = 0.5, 1, 1.5, 2, 2.5 and 3.

    o ppm, kppm, dppm, slrm, mppm
    These model-fitting functions now handle the case where the
    point pattern is an 'NAobject'.

    o simulate.ppm, rmh.ppm, rmhmodel.ppm
    New argument 'newdata' allows the model to be simulated
    in a completely different window, or using different covariate values.

    o simulate.ppm, rmh.ppm, rmhmodel.ppm
    A warning or error occurs if there are too many NA values
    of the model trend in the desired window.

BUG FIXES

    o predict.mppm
    Crashed if a covariate contained NA values and
    the model included random effects.
    [Spotted by Andrea Gilardi.]
    Fixed.

        CHANGES IN spatstat.model VERSION 3.4-0

OVERVIEW

    o Extensions to ROC curves.

NEW FUNCTIONS

    o addROC
    Calculate partial ROC curve for adding a new covariate to a model.

    o dropROC
    Calculate partial ROC curve for removing a covariate from a model.

    o dropply, addapply
    Consider all single-variable additions or deletions in a model.

SIGNIFICANT USER-VISIBLE CHANGES

    o mppm
    Recognises argument 'quad.args'.

    o roc.ppm, roc.slrm, roc.kppm
    New argument 'baseline' allows calculation of ROC relative to a baseline.

    o roc.ppm, roc.slrm, roc.kppm
    New argument 'method' determines the estimation method.
    New options include a kernel smoothing estimate and a monotone estimate.

    o roc.ppm, roc.slrm, roc.kppm
    New arguments 'CI', 'alpha' for calculating confidence intervals.

    o roc.ppm, roc.slrm, roc.kppm
    New argument 'subset' specifies a sub-region of the spatial domain
    in which the ROC should be calculated.

    o roc.ppm, roc.slrm, roc.kppm
    New argument 'leaveoneout' specifies whether to use leave-one-out estimates
    of intensity at the data points.

        CHANGES IN spatstat.model VERSION 3.3-6

OVERVIEW

    o Simulation of 'zclustermodel' objects.

    o Documentation improvements and bug fixes.

NEW FUNCTIONS

    o simulate.zclustermodel
    Method for simulating objects of class 'zclustermodel'.

BUG FIXES

    o intensity.zclustermodel
    Always returned the value zero.
    Fixed.

        CHANGES IN spatstat.model VERSION 3.3-5

OVERVIEW

    o We thank John Ginos for contributions.

    o Bug fix.

BUG FIXES

    o simulate.kppm
    Argument saveLambda=TRUE was ignored for nonstationary LGCP models.
    [Spotted by John Ginos.]
    Fixed.


To generate a diff of this commit:
cvs rdiff -u -r1.3 -r1.4 pkgsrc/geography/R-spatstat.model/Makefile \
    pkgsrc/geography/R-spatstat.model/distinfo

Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.

Modified files:

Index: pkgsrc/geography/R-spatstat.model/Makefile
diff -u pkgsrc/geography/R-spatstat.model/Makefile:1.3 pkgsrc/geography/R-spatstat.model/Makefile:1.4
--- pkgsrc/geography/R-spatstat.model/Makefile:1.3      Sat Feb  8 09:28:38 2025
+++ pkgsrc/geography/R-spatstat.model/Makefile  Fri Jun 12 13:59:24 2026
@@ -1,7 +1,7 @@
-# $NetBSD: Makefile,v 1.3 2025/02/08 09:28:38 mef Exp $
+# $NetBSD: Makefile,v 1.4 2026/06/12 13:59:24 mef Exp $
 
 R_PKGNAME=     spatstat.model
-R_PKGVER=      3.3-4
+R_PKGVER=      3.7-1
 CATEGORIES=    geography
 
 MAINTAINER=    pkgsrc-users%NetBSD.org@localhost
Index: pkgsrc/geography/R-spatstat.model/distinfo
diff -u pkgsrc/geography/R-spatstat.model/distinfo:1.3 pkgsrc/geography/R-spatstat.model/distinfo:1.4
--- pkgsrc/geography/R-spatstat.model/distinfo:1.3      Sat Feb  8 09:28:38 2025
+++ pkgsrc/geography/R-spatstat.model/distinfo  Fri Jun 12 13:59:24 2026
@@ -1,5 +1,5 @@
-$NetBSD: distinfo,v 1.3 2025/02/08 09:28:38 mef Exp $
+$NetBSD: distinfo,v 1.4 2026/06/12 13:59:24 mef Exp $
 
-BLAKE2s (R/spatstat.model_3.3-4.tar.gz) = 5b6751b368a5e4061d1df53079bf92bd1da5f6659e0c38f16bf59b50ed08adf5
-SHA512 (R/spatstat.model_3.3-4.tar.gz) = b1163223ac0e4abd048df02162b4199d84ef032dffb839304c8818fcb69cb9e46d36a8af45b320c9282dab0b7b9933fec2b0c64c416355e6fae50d836590c214
-Size (R/spatstat.model_3.3-4.tar.gz) = 807732 bytes
+BLAKE2s (R/spatstat.model_3.7-1.tar.gz) = 91ff19eda4019406fd10daa169c9fbb218cff05728cd34c360160248639f155e
+SHA512 (R/spatstat.model_3.7-1.tar.gz) = f5e5b4b70a8c3af9368294f855e37468eef91d4c789c43f257d0ee5ea6fe947a3b11924da71a6e8a028154a158a8443cfa34e91ed5dc74aaec35addbe4c76b37
+Size (R/spatstat.model_3.7-1.tar.gz) = 864211 bytes