lppm recognises seg and tp

Author: baddstats

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: spatstat.linnet
-Version: 3.4-0.019
-Date: 2026-01-17
+Version: 3.4-0.020
+Date: 2026-01-19
 Title: Linear Networks Functionality of the 'spatstat' Family
 Authors@R: c(person("Adrian", "Baddeley", 
                     role = c("aut", "cre", "cph"),
@@ -25,7 +25,7 @@ Authors@R: c(person("Adrian", "Baddeley",
 	     person("Ottmar", "Cronie",
 	            role = "ctb"))
 Maintainer: Adrian Baddeley <Adrian.Baddeley@curtin.edu.au>
-Depends: R (>= 3.5.0), stats, graphics, grDevices, methods, utils, spatstat.data (>= 3.1-9), spatstat.univar (>= 3.1-6), spatstat.geom (>= 3.6-1.022), spatstat.random (>= 3.4-3), spatstat.explore (>= 3.6-0), spatstat.model (>= 3.5-0.012)
+Depends: R (>= 3.5.0), stats, graphics, grDevices, methods, utils, spatstat.data (>= 3.1-9), spatstat.univar (>= 3.1-6), spatstat.geom (>= 3.6-1.026), spatstat.random (>= 3.4-3), spatstat.explore (>= 3.6-0), spatstat.model (>= 3.5-0.023)
 Imports: Matrix, spatstat.utils (>= 3.2-1), spatstat.sparse (>= 3.1)
 Suggests: goftest, locfit, spatstat (>= 3.4)
 Description: Defines types of spatial data on a linear network

NEWS

@@ -1,5 +1,5 @@
 
-        CHANGES IN spatstat.linnet VERSION 3.4-0.019
+        CHANGES IN spatstat.linnet VERSION 3.4-0.020
 
 OVERVIEW
 

R/diaglppm.R

@@ -4,7 +4,7 @@
 ## Makes diagnostic plots based on residuals or energy weights
 ## for a point process model on a network
 ##
-## $Revision: 1.4 $ $Date: 2025/11/24 04:29:25 $
+## $Revision: 1.5 $ $Date: 2026/01/19 05:49:54 $
 ##
 
 diagnose.lppm <- function(object, ..., type="raw", which="all", 
@@ -139,7 +139,7 @@ diagLppmEngine <- function(object, ..., type="eem", typename, opt,
   W <- Window(Q)
 
   ## quadrature points on the network
-  U <- attr(Q, "plekken") %orifnull% lpp(U2D, L)
+  U <- attr(Q, "situ") %orifnull% lpp(U2D, L)
 
   ## 
   ppmfit <- as.ppm(object)

R/linequad.R

@@ -1,7 +1,7 @@
 #
 # linequad.R
 #
-#  $Revision: 1.32 $ $Date: 2026/01/09 01:35:25 $
+#  $Revision: 1.33 $ $Date: 2026/01/19 05:50:01 $
 #
 # create quadscheme for a pattern of points lying *on* line segments
 
@@ -350,8 +350,8 @@ linequad <- function(X, Y, ..., eps=NULL, nd=1000, random=FALSE) {
                 weight = list(method=wmethod))
   ## make quad scheme
   Qout <- quad(as.ppp(DAT), as.ppp(DUM), c(wdat, wdum), param=param)
-  ## add information
-  attr(Qout, "plekken") <- superimpose(DAT, DUM, L=L)
+  ## add information: point locations 'in situ'
+  attr(Qout, "situ") <- superimpose(DAT, DUM, L=L)
   ## 
   return(Qout)
 }

R/lppm.R

@@ -3,7 +3,7 @@
 #
 #  Point process models on a linear network
 #
-#  $Revision: 1.68 $   $Date: 2025/12/19 07:50:29 $
+#  $Revision: 1.69 $   $Date: 2026/01/19 05:49:47 $
 #
 
 lppm <- function(X, ...) {
@@ -126,12 +126,29 @@ predict.lppm <- local({
     X <- object$X
     fit <- object$fit
     L <- as.linnet(X)
-    
+
+    ## functions to evaluate the local covariates
+    LocalCoords <- list(seg = linfun(function(x,y,seg,tp) { seg }, L),
+                        tp  = linfun(function(x,y,seg,tp) { seg }, L))
+
     if(!is.null(locations)) {
-      #' locations given; return a vector/matrix of predicted values
-      if(is.lpp(locations)) locations <- as.ppp(locations)
+      ## determine whether 'locations' includes local coordinates
+      if(is.data.frame(locations)) {
+        ## data frame of spatial locations
+        gotlocal <- all(c("seg", "tp") %in% names(locations))
+      } else if(is.lpp(locations)) {
+        ## point pattern on network
+        gotlocal <- TRUE
+        loci <- locations
+        locations <- as.ppp(locations)
+        attr(locations, "situ") <- loci
+      } else {
+        ## other spatial data
+        gotlocal <- FALSE
+      }
       values <- predict(fit, locations=locations, covariates=covariates,
-                        type=type, se=se, new.coef=new.coef)
+                        type=type, se=se, new.coef=new.coef,
+                        extracovariates=if(!gotlocal) LocalCoords else NULL)
       return(values)
     }
 
@@ -157,7 +174,8 @@ predict.lppm <- local({
     if(!is.multitype(fit)) {
       #' unmarked
       values <- predict(fit, locations=projloc, covariates=covariates,
-                        type=type, se=se, new.coef=new.coef)
+                        type=type, se=se, new.coef=new.coef,
+                        extracovariates=LocalCoords)
       if(!se) {
         out <- putvalues(values, lineimage, pixelcentres, projdata, L)
       } else {
@@ -175,7 +193,8 @@ predict.lppm <- local({
         markk <- factor(lev[k], levels=lev)
         locnk <- cbind(projloc, data.frame(marks=markk))
         values <- predict(fit, locations=locnk, covariates=covariates,
-                          type=type, se=se, new.coef=new.coef)
+                          type=type, se=se, new.coef=new.coef,
+                          extracovariates=LocalCoords)
         if(!se) {
           out[[k]] <- putvalues(values, lineimage, pixelcentres, projdata, L)
         } else {
@@ -512,6 +531,6 @@ eem.lppm <- function(fit, ...) {
 
 residuals.lppm <- function(object, type="raw", ...) {
   res <- residuals(as.ppm(object), type=type, ...)
-  attr(res, "plekken") <- attr(quad.ppm(object), "plekken")
+  attr(res, "situ") <- attr(quad.ppm(object), "situ")
   return(res)
 }

R/lurklppm.R

@@ -51,9 +51,14 @@ lurking.lpp <- lurking.lppm <- function(object, covariate,
     co <- cl$covariate
     covname <- if(is.name(co)) as.character(co) else
                if(is.expression(co)) format(co[[1]]) else 
-               if(is.character(co) && 
-                  length(co) == 1 &&
-                  co %in% c("x", "y")) paste(co, "coordinate") else NULL
+               if(is.character(co) && length(co) == 1) {
+                 switch(co,
+                        x = "x coordinate",
+                        y = "y coordinate",
+                        seg = "segment number",
+                        tp = "tp coordinate",
+                        NULL)
+               } else NULL
   }
 
   #' spatial covariates
@@ -107,6 +112,8 @@ lurking.lpp <- lurking.lppm <- function(object, covariate,
   quadpoints <- union.quad(Q)
   Z <- is.data(Q)
   wts <- w.quad(Q)
+  ## positions on the network
+  situ <- attr(Q, "situ")
   ## subset of quadrature points used to fit model
   subQset <- getglmsubset(object2D)
   if(is.null(subQset)) subQset <- rep.int(TRUE, n.quad(Q))
@@ -115,30 +122,39 @@ lurking.lpp <- lurking.lppm <- function(object, covariate,
   #' trap case where covariate = "<name>"
   if(is.character(covariate) && (length(covariate) == 1)) {
     #' covariate is a single string
-    is.cartesian <- covariate %in% c("x", "y")
-    if(!is.cartesian) {
+    is.coordinate <- covariate %in% c("x", "y", "seg", "tp")
+    if(!is.coordinate) {
       #' not a reserved name; convert to an expression and evaluate later
       covariate <- str2expression(covariate)
     }
   } else {
-    is.cartesian <- FALSE
+    is.coordinate <- FALSE
   }
 
   #################################################################
   ## compute the covariate
   covunits <- NULL
-  if(is.cartesian) {
-    #' covariate is name of cartesian coordinate
+  if(is.coordinate) {
+    #' covariate is name of a spatial or local coordinate
     switch(covariate,
            x = {
              covvalues <- quadpoints$x
              covrange <- Frame(quadpoints)$xrange
+             covunits <- unitname(quadpoints)
            },
            y = {
              covvalues <- quadpoints$y
              covrange <- Frame(quadpoints)$yrange
+             covunits <- unitname(quadpoints)
+           },
+           seg = {
+             covvalues <- coords(situ)$seg
+             covrange <- range(covvalues)
+           },
+           tp = {
+             covvalues <- coords(situ)$tp
+             covrange <- c(0,1)
            })
-    covunits <- unitname(quadpoints)
   } else if(is.im(covariate)) {
     covvalues <- covariate[quadpoints, drop=FALSE]
     covrange <- internal$covrange %orifnull% range(covariate, finite=TRUE)

inst/doc/packagesizes.txt

@@ -42,4 +42,4 @@ date version nhelpfiles nobjects ndatasets Rlines srclines
 "2025-07-24" "3.3-1" 150 328 0 13326 3270
 "2025-09-24" "3.3-2" 150 328 0 13356 3270
 "2025-11-29" "3.4-0" 160 370 0 14990 3270
-"2026-01-17" "3.4-0.019" 162 374 0 15471 3270
+"2026-01-19" "3.4-0.020" 162 374 0 15506 3270

inst/info/packagesizes.txt

@@ -42,4 +42,4 @@ date version nhelpfiles nobjects ndatasets Rlines srclines
 "2025-07-24" "3.3-1" 150 328 0 13326 3270
 "2025-09-24" "3.3-2" 150 328 0 13356 3270
 "2025-11-29" "3.4-0" 160 370 0 14990 3270
-"2026-01-17" "3.4-0.019" 162 374 0 15471 3270
+"2026-01-19" "3.4-0.020" 162 374 0 15506 3270

man/lppm.Rd

@@ -101,20 +101,10 @@ lppm(X, ...)
   lppm(X ~ marks * x)
 }
 \references{
-  Ang, Q.W. (2010)
-  \emph{Statistical methodology for events on a network}.
-  Master's thesis, School of Mathematics and Statistics, University of
-  Western Australia.
-  
-  Ang, Q.W., Baddeley, A. and Nair, G. (2012)
-  Geometrically corrected second-order analysis of 
-  events on a linear network, with applications to
-  ecology and criminology.
-  \emph{Scandinavian Journal of Statistics} \bold{39}, 591--617.
-
-  McSwiggan, G., Nair, M.G. and Baddeley, A. (2012)
-  Fitting Poisson point process models to events 
-  on a linear network. Manuscript in preparation.
+  McSwiggan, G. (2019)  
+  Spatial point process methods for linear networks
+  with applications to road accident analysis.
+  PhD thesis, University of Western Australia.
 }
 \keyword{spatial}
 \keyword{models}

man/predict.lppm.Rd

@@ -24,8 +24,9 @@
   }
   \item{locations}{
     Optional. Locations at which predictions should be computed.
-    Either a data frame with two columns of coordinates,
-    or a binary image mask.
+    Either a point pattern (class \code{"lpp"} or \code{"ppp"}),
+    a data frame containing spatial coordinates,
+    or a binary image mask, or a pixel image.
   }
   \item{covariates}{
     Values of external covariates required by the model.
@@ -78,11 +79,14 @@
     (If the model is multitype, the result is a list of such pixel
     images, one for each possible type of point.)
     \item
-    If \code{locations} is a data frame, the result is a 
+    If \code{locations} is a data frame containing
+    spatial coordinates \code{x} and \code{y}, and/or 
+    local coordinates \code{seg} and \code{tp}, the result is a 
     numeric vector of predicted values at the locations specified by
     the data frame.
     \item
-    If \code{locations} is a binary mask, the result is a pixel image
+    If \code{locations} is a binary mask or pixel image,
+    the result is a pixel image
     with predicted values computed at the pixels of the mask.
     (If the model is multitype, the result is a list of such pixel
     images, one for each possible type of point.)