spatstat (version 1.25-4)

spatstat-package: The Spatstat Package

Description

This is a summary of the features of spatstat, a package in R for the statistical analysis of spatial point patterns.

Arguments

Getting Started

For a quick introduction to spatstat, see the package vignette Getting started with spatstat installed with spatstat. (To see this document online, start R, type help.start() to open the help browser, and navigate to Packages > spatstat > Vignettes).

For a complete 2-day course on using spatstat, see the workshop notes by Baddeley (2010), available on the internet.

Type demo(spatstat) for a demonstration of the package's capabilities. Type demo(data) to see all the datasets available in the package. For information about handling data in shapefiles, see the Vignette Handling shapefiles in the spatstat package installed with spatstat.

To learn about spatial point process methods, see the short book by Diggle (2003) and the handbook Gelfand et al (2010).

Updates

New versions of spatstat are produced about once a month. Users are advised to update their installation of spatstat regularly. Type latest.news() to read the news documentation about changes to the current installed version of spatstat. Type news(package="spatstat") to read news documentation about all previous versions of the package.

FUNCTIONS AND DATASETS

Following is a summary of the main functions and datasets in the spatstat package. Alternatively an alphabetical list of all functions and datasets is available by typing library(help=spatstat).

For further information on any of these, type help(name) where name is the name of the function or dataset.

CONTENTS:

ll{ I. Creating and manipulating data II. Exploratory Data Analysis III. Model fitting (cluster models) IV. Model fitting (Poisson and Gibbs models) V. Model fitting (spatial logistic regression) VI. Simulation VII. Tests and diagnostics VIII. Documentation }

I. CREATING AND MANIPULATING DATA

Types of spatial data:

The main types of spatial data supported by spatstat are:

ll{ ppp point pattern owin window (spatial region) im pixel image psp line segment pattern tess tessellation pp3 three-dimensional point pattern ppx point pattern in any number of dimensions lpp point pattern on a linear network }

To create a point pattern: ll{ ppp create a point pattern from $(x,y)$ and window information ppp(x, y, xlim, ylim) for rectangular window ppp(x, y, poly) for polygonal window ppp(x, y, mask) for binary image window as.ppp convert other types of data to a ppp object clickppp interactively add points to a plot marks<-, %mark% attach/reassign marks to a point pattern } To simulate a random point pattern: ll{ runifpoint generate $n$ independent uniform random points rpoint generate $n$ independent random points rmpoint generate $n$ independent multitype random points rpoispp simulate the (in)homogeneous Poisson point process rmpoispp simulate the (in)homogeneous multitype Poisson point process runifdisc generate $n$ independent uniform random points in disc rstrat stratified random sample of points rsyst systematic random sample of points rjitter apply random displacements to points in a pattern rMaternI simulate the Mat'ern Model I inhibition process rMaternII simulate the Mat'ern Model II inhibition process rSSI simulate Simple Sequential Inhibition process rStrauss simulate Strauss process (perfect simulation) rHardcore simulate Hard Core process (perfect simulation) rDiggleGratton simulate Diggle-Gratton process (perfect simulation) rDGS simulate Diggle-Gates-Stibbard process (perfect simulation) rNeymanScott simulate a general Neyman-Scott process rPoissonCluster simulate a general Neyman-Scott process rNeymanScott simulate a general Neyman-Scott process rMatClust simulate the Mat'ern Cluster process rThomas simulate the Thomas process rGaussPoisson simulate the Gauss-Poisson cluster process rCauchy simulate Neyman-Scott Cauchy cluster process rVarGamma simulate Neyman-Scott Variance Gamma cluster process rthin random thinning rcell simulate the Baddeley-Silverman cell process rmh simulate Gibbs point process using Metropolis-Hastings simulate.ppm simulate Gibbs point process using Metropolis-Hastings runifpointOnLines generate $n$ random points along specified line segments rpoisppOnLines generate Poisson random points along specified line segments }

To randomly change an existing point pattern: ll{ rshift random shifting of points rjitter apply random displacements to points in a pattern rthin random thinning rlabel random (re)labelling of a multitype point pattern quadratresample block resampling }

Standard point pattern datasets:

Datasets in spatstat are lazy-loaded, so you can simply type the name of the dataset to use it; there is no need to type data(amacrine) etc.

Type demo(data) to see a display of all the datasets installed with the package. ll{ amacrine Austin Hughes' rabbit amacrine cells anemones Upton-Fingleton sea anemones data ants Harkness-Isham ant nests data bei Tropical rainforest trees betacells Waessle et al. cat retinal ganglia data bramblecanes Bramble Canes data bronzefilter Bronze Filter Section data cells Crick-Ripley biological cells data chicago Chicago street crimes chorley Chorley-Ribble cancer data copper Berman-Huntington copper deposits data demopat Synthetic point pattern finpines Finnish Pines data flu Influenza virus proteins gorillas Gorilla nest sites hamster Aherne's hamster tumour data humberside North Humberside childhood leukaemia data japanesepines Japanese Pines data lansing Lansing Woods data longleaf Longleaf Pines data murchison Murchison gold deposits nbfires New Brunswick fires data nztrees Mark-Esler-Ripley trees data osteo Osteocyte lacunae (3D, replicated) ponderosa Getis-Franklin ponderosa pine trees data redwood Strauss-Ripley redwood saplings data redwoodfull Strauss redwood saplings data (full set) residualspaper Data from Baddeley et al (2005) shapley Galaxies in an astronomical survey simdat Simulated point pattern (inhomogeneous, with interaction) spruces Spruce trees in Saxonia swedishpines Strand-Ripley swedish pines data urkiola Urkiola Woods data }

To manipulate a point pattern:

ll{ plot.ppp plot a point pattern (e.g. plot(X)) iplot plot a point pattern interactively [.ppp extract or replace a subset of a point pattern pp[subset] or pp[subwindow] superimpose combine several point patterns by.ppp apply a function to sub-patterns of a point pattern cut.ppp classify the points in a point pattern unmark remove marks npoints count the number of points coords extract coordinates, change coordinates marks extract marks, change marks or attach marks split.ppp divide pattern into sub-patterns rotate rotate pattern shift translate pattern flipxy swap $x$ and $y$ coordinates periodify make several translated copies affine apply affine transformation density.ppp kernel smoothing of point pattern smooth.ppp smooth the marks attached to points sharpen.ppp data sharpening identify.ppp interactively identify points unique.ppp remove duplicate points duplicated.ppp determine which points are duplicates dirichlet compute Dirichlet-Voronoi tessellation delaunay compute Delaunay triangulation convexhull compute convex hull discretise discretise coordinates pixellate.ppp approximate point pattern by pixel image as.im.ppp approximate point pattern by pixel image } See spatstat.options to control plotting behaviour. To create a window:

An object of class "owin" describes a spatial region (a window of observation).

ll{ owin Create a window object owin(xlim, ylim) for rectangular window owin(poly) for polygonal window owin(mask) for binary image window as.owin Convert other data to a window object square make a square window disc make a circular window ripras Ripley-Rasson estimator of window, given only the points convexhull compute convex hull of something letterR polygonal window in the shape of the Rlogo }

To manipulate a window:

ll{ plot.owin plot a window. plot(W) bounding.box Find a tight bounding box for the window erosion erode window by a distance r dilation dilate window by a distance r closing close window by a distance r opening open window by a distance r border difference between window and its erosion/dilation complement.owin invert (swap inside and outside) simplify.owin approximate a window by a simple polygon rotate rotate window flipxy swap $x$ and $y$ coordinates shift translate window periodify make several translated copies affine apply affine transformation }

Digital approximations:

ll{ as.mask Make a discrete pixel approximation of a given window as.im.owin convert window to pixel image pixellate.owin convert window to pixel image commonGrid find common pixel grid for windows nearest.raster.point map continuous coordinates to raster locations raster.x raster x coordinates raster.y raster y coordinates as.polygonal convert pixel mask to polygonal window } See spatstat.options to control the approximation

Geometrical computations with windows:

ll{ intersect.owin intersection of two windows union.owin union of two windows setminus.owin set subtraction of two windows inside.owin determine whether a point is inside a window area.owin compute area perimeter compute perimeter length diameter.owin compute diameter incircle find largest circle inside a window connected find connected components of window eroded.areas compute areas of eroded windows dilated.areas compute areas of dilated windows bdist.points compute distances from data points to window boundary bdist.pixels compute distances from all pixels to window boundary bdist.tiles boundary distance for each tile in tessellation distmap.owin distance transform image distfun.owin distance transform centroid.owin compute centroid (centre of mass) of window is.subset.owin determine whether one window contains another is.convex determine whether a window is convex convexhull compute convex hull as.mask pixel approximation of window as.polygonal polygonal approximation of window setcov spatial covariance function of window }

Pixel images: An object of class "im" represents a pixel image. Such objects are returned by some of the functions in spatstat including Kmeasure, setcov and density.ppp. ll{ im create a pixel image as.im convert other data to a pixel image pixellate convert other data to a pixel image as.matrix.im convert pixel image to matrix as.data.frame.im convert pixel image to data frame plot.im plot a pixel image on screen as a digital image contour.im draw contours of a pixel image persp.im draw perspective plot of a pixel image rgbim create colour-valued pixel image hsvim create colour-valued pixel image [.im extract a subset of a pixel image [<-.im replace a subset of a pixel image shift.im apply vector shift to pixel image X print very basic information about image X summary(X) summary of image X hist.im histogram of image mean.im mean pixel value of image integral.im integral of pixel values quantile.im quantiles of image cut.im convert numeric image to factor image is.im test whether an object is a pixel image interp.im interpolate a pixel image blur apply Gaussian blur to image connected find connected components compatible.im test whether two images have compatible dimensions harmonise.im make images compatible commonGrid find a common pixel grid for images eval.im evaluate any expression involving images scaletointerval rescale pixel values zapsmall.im set very small pixel values to zero levelset level set of an image solutionset region where an expression is true imcov spatial covariance function of image }

Line segment patterns

An object of class "psp" represents a pattern of straight line segments. ll{ psp create a line segment pattern as.psp convert other data into a line segment pattern is.psp determine whether a dataset has class "psp" plot.psp plot a line segment pattern print.psp print basic information summary.psp print summary information [.psp extract a subset of a line segment pattern as.data.frame.psp convert line segment pattern to data frame marks.psp extract marks of line segments marks<-.psp assign new marks to line segments unmark.psp delete marks from line segments midpoints.psp compute the midpoints of line segments endpoints.psp extract the endpoints of line segments lengths.psp compute the lengths of line segments angles.psp compute the orientation angles of line segments superimpose combine several line segment patterns flipxy swap $x$ and $y$ coordinates rotate.psp rotate a line segment pattern shift.psp shift a line segment pattern periodify make several shifted copies affine.psp apply an affine transformation pixellate.psp approximate line segment pattern by pixel image as.mask.psp approximate line segment pattern by binary mask distmap.psp compute the distance map of a line segment pattern distfun.psp compute the distance map of a line segment pattern density.psp kernel smoothing of line segments selfcrossing.psp find crossing points between line segments crossing.psp find crossing points between two line segment patterns nncross find distance to nearest line segment from a given point nearestsegment find line segment closest to a given point project2segment find location along a line segment closest to a given point pointsOnLines generate points evenly spaced along line segment rpoisline generate a realisation of the Poisson line process inside a window rlinegrid generate a random array of parallel lines through a window }

Tessellations

An object of class "tess" represents a tessellation.

ll{ tess create a tessellation quadrats create a tessellation of rectangles as.tess convert other data to a tessellation plot.tess plot a tessellation tiles extract all the tiles of a tessellation [.tess extract some tiles of a tessellation [<-.tess change some tiles of a tessellation intersect.tess intersect two tessellations or restrict a tessellation to a window chop.tess subdivide a tessellation by a line dirichlet compute Dirichlet-Voronoi tessellation of points delaunay compute Delaunay triangulation of points rpoislinetess generate tessellation using Poisson line process tile.areas area of each tile in tessellation bdist.tiles boundary distance for each tile in tessellation }

Three-dimensional point patterns

An object of class "pp3" represents a three-dimensional point pattern in a rectangular box. The box is represented by an object of class "box3".

ll{ pp3 create a 3-D point pattern plot.pp3 plot a 3-D point pattern coords extract coordinates as.hyperframe extract coordinates unitname.pp3 name of unit of length npoints count the number of points runifpoint3 generate uniform random points in 3-D rpoispp3 generate Poisson random points in 3-D envelope.pp3 generate simulation envelopes for 3-D pattern box3 create a 3-D rectangular box as.box3 convert data to 3-D rectangular box unitname.box3 name of unit of length diameter.box3 diameter of box volume.box3 volume of box shortside.box3 shortest side of box eroded.volumes volumes of erosions of box }

Multi-dimensional space-time point patterns

An object of class "ppx" represents a point pattern in multi-dimensional space and/or time.

ll{ ppx create a multidimensional space-time point pattern coords extract coordinates as.hyperframe extract coordinates unitname.ppx name of unit of length npoints count the number of points runifpointx generate uniform random points rpoisppx generate Poisson random points boxx define multidimensional box diameter.boxx diameter of box volume.boxx volume of box shortside.boxx shortest side of box eroded.volumes.boxx volumes of erosions of box } Point patterns on a linear network

An object of class "linnet" represents a linear network (for example, a road network).

ll{ linnet create a linear network clickjoin interactively join vertices in network simplenet simple example of network lineardisc disc in a linear network methods.linnet methods for linnet objects } An object of class "lpp" represents a point pattern on a linear network (for example, road accidents on a road network). ll{ lpp create a point pattern on a linear network methods.lpp methods for lpp objects rpoislpp simulate Poisson points on linear network runiflpp simulate random points on a linear network chicago Chicago street crime data } Hyperframes

A hyperframe is like a data frame, except that the entries may be objects of any kind.

ll{ hyperframe create a hyperframe as.hyperframe convert data to hyperframe plot.hyperframe plot hyperframe with.hyperframe evaluate expression using each row of hyperframe cbind.hyperframe combine hyperframes by columns rbind.hyperframe combine hyperframes by rows as.data.frame.hyperframe convert hyperframe to data frame } Layered objects

A layered object represents data that should be plotted in successive layers, for example, a background and a foreground.

ll{ layered create layered object plot.layered plot layered object }

II. EXPLORATORY DATA ANALYSIS

Inspection of data: ll{ summary(X) print useful summary of point pattern X X print basic description of point pattern X any(duplicated(X)) check for duplicated points in pattern X istat(X) Interactive exploratory analysis }

Classical exploratory tools: ll{ clarkevans Clark and Evans aggregation index fryplot Fry plot miplot Morishita Index plot }

Modern exploratory tools: ll{ nnclean Byers-Raftery feature detection sharpen.ppp Choi-Hall data sharpening rhohat Smoothing estimate of covariate effect }

Summary statistics for a point pattern: ll{ quadratcount Quadrat counts Fest empty space function $F$ Gest nearest neighbour distribution function $G$ Jest $J$-function $J = (1-G)/(1-F)$ Kest Ripley's $K$-function Lest Besag $L$-function Tstat Third order $T$-function allstats all four functions $F$, $G$, $J$, $K$ pcf pair correlation function Kinhom $K$ for inhomogeneous point patterns Linhom $L$ for inhomogeneous point patterns pcfinhom pair correlation for inhomogeneous patterns localL Getis-Franklin neighbourhood density function localK neighbourhood K-function localpcf local pair correlation function localKinhom local $K$ for inhomogeneous point patterns localLinhom local $L$ for inhomogeneous point patterns localpcfinhom local pair correlation for inhomogeneous patterns Kest.fft fast $K$-function using FFT for large datasets Kmeasure reduced second moment measure envelope simulation envelopes for a summary function varblock variances and confidence intervals for a summary function }

Related facilities: ll{ plot.fv plot a summary function eval.fv evaluate any expression involving summary functions eval.fasp evaluate any expression involving an array of functions with.fv evaluate an expression for a summary function smooth.fv apply smoothing to a summary function nndist nearest neighbour distances nnwhich find nearest neighbours pairdist distances between all pairs of points crossdist distances between points in two patterns nncross nearest neighbours between two point patterns exactdt distance from any location to nearest data point distmap distance map image distfun distance map function density.ppp kernel smoothed density smooth.ppp spatial interpolation of marks relrisk kernel estimate of relative risk sharpen.ppp data sharpening rknn theoretical distribution of nearest neighbour distance }

Summary statistics for a multitype point pattern: A multitype point pattern is represented by an object X of class "ppp" such that marks(X) is a factor. ll{ relrisk kernel estimation of relative risk scan.test spatial scan test of elevated risk Gcross,Gdot,Gmulti multitype nearest neighbour distributions $G_{ij}, G_{i\bullet}$ Kcross,Kdot, Kmulti multitype $K$-functions $K_{ij}, K_{i\bullet}$ Lcross,Ldot multitype $L$-functions $L_{ij}, L_{i\bullet}$ Jcross,Jdot,Jmulti multitype $J$-functions $J_{ij}, J_{i\bullet}$ pcfcross multitype pair correlation function $g_{ij}$ pcfdot multitype pair correlation function $g_{i\bullet}$ markconnect marked connection function $p_{ij}$ alltypes estimates of the above for all $i,j$ pairs Iest multitype $I$-function Kcross.inhom,Kdot.inhom inhomogeneous counterparts of Kcross, Kdot Lcross.inhom,Ldot.inhom inhomogeneous counterparts of Lcross, Ldot pcfcross.inhom,pcfdot.inhom inhomogeneous counterparts of pcfcross, pcfdot }

Summary statistics for a marked point pattern: A marked point pattern is represented by an object X of class "ppp" with a component X$marks. The entries in the vector X$marks may be numeric, complex, string or any other atomic type. For numeric marks, there are the following functions: ll{ markmean smoothed local average of marks markvar smoothed local variance of marks markcorr mark correlation function markvario mark variogram markcorrint mark correlation integral Emark mark independence diagnostic $E(r)$ Vmark mark independence diagnostic $V(r)$ nnmean nearest neighbour mean index nnvario nearest neighbour mark variance index } For marks of any type, there are the following: ll{ Gmulti multitype nearest neighbour distribution Kmulti multitype $K$-function Jmulti multitype $J$-function } Alternatively use cut.ppp to convert a marked point pattern to a multitype point pattern.

Programming tools: ll{ applynbd apply function to every neighbourhood in a point pattern markstat apply function to the marks of neighbours in a point pattern marktable tabulate the marks of neighbours in a point pattern pppdist find the optimal match between two point patterns }

Summary statistics for a point pattern on a linear network:

These are for point patterns on a linear network (class lpp). ll{ linearK $K$ function on linear network linearKinhom inhomogeneous $K$ function on linear network linearpcf pair correlation function on linear network linearpcfinhom inhomogeneous pair correlation on linear network }

Related facilities: ll{ pairdist.lpp shortest path distances envelope.lpp simulation envelopes rpoislpp simulate Poisson points on linear network runiflpp simulate random points on a linear network } It is also possible to fit point process models to lpp objects. See Section IV. Summary statistics for a three-dimensional point pattern:

These are for 3-dimensional point pattern objects (class pp3).

ll{ F3est empty space function $F$ G3est nearest neighbour function $G$ K3est $K$-function pcf3est pair correlation function }

Related facilities: ll{ envelope.pp3 simulation envelopes pairdist.pp3 distances between all pairs of points crossdist.pp3 distances between points in two patterns nndist.pp3 nearest neighbour distances nnwhich.pp3 find nearest neighbours }

Computations for multi-dimensional point pattern:

These are for multi-dimensional space-time point pattern objects (class ppx).

ll{ pairdist.ppx distances between all pairs of points crossdist.ppx distances between points in two patterns nndist.ppx nearest neighbour distances nnwhich.ppx find nearest neighbours }

Summary statistics for random sets: These work for point patterns (class ppp), line segment patterns (class psp) or windows (class owin). ll{ Hest spherical contact distribution $H$ Gfox Foxall $G$-function Jfox Foxall $J$-function }

III. MODEL FITTING (CLUSTER MODELS)

Cluster process models (with homogeneous or inhomogeneous intensity) and Cox processes can be fitted by the function kppm. Its result is an object of class "kppm". The fitted model can be printed, plotted, predicted, simulated and updated.

ll{ kppm Fit model plot.kppm Plot the fitted model predict.kppm Compute fitted intensity update.kppm Update the model simulate.kppm Generate simulated realisations vcov.kppm Variance-covariance matrix of coefficients Kmodel.kppm $K$ function of fitted model pcfmodel.kppm Pair correlation of fitted model } The theoretical models can also be simulated, for any choice of parameter values, using rThomas, rMatClust, rCauchy, rVarGamma, and rLGCP. Lower-level fitting functions include:

ll{ lgcp.estK fit a log-Gaussian Cox process model lgcp.estpcf fit a log-Gaussian Cox process model thomas.estK fit the Thomas process model thomas.estpcf fit the Thomas process model matclust.estK fit the Matern Cluster process model matclust.estpcf fit the Matern Cluster process model cauchy.estK fit a Neyman-Scott Cauchy cluster process cauchy.estpcf fit a Neyman-Scott Cauchy cluster process vargamma.estK fit a Neyman-Scott Variance Gamma process vargamma.estpcf fit a Neyman-Scott Variance Gamma process mincontrast low-level algorithm for fitting models by the method of minimum contrast }

IV. MODEL FITTING (POISSON AND GIBBS MODELS)

Types of models Poisson point processes are the simplest models for point patterns. A Poisson model assumes that the points are stochastically independent. It may allow the points to have a non-uniform spatial density. The special case of a Poisson process with a uniform spatial density is often called Complete Spatial Randomness. Poisson point processes are included in the more general class of Gibbs point process models. In a Gibbs model, there is interaction or dependence between points. Many different types of interaction can be specified. For a detailed explanation of how to fit Poisson or Gibbs point process models to point pattern data using spatstat, see Baddeley and Turner (2005b) or Baddeley (2008). To fit a Poison or Gibbs point process model:

Model fitting in spatstat is performed mainly by the function ppm. Its result is an object of class "ppm". Here are some examples, where X is a point pattern (class "ppp"): ll{ command model ppm(X) Complete Spatial Randomness ppm(X, ~1) Complete Spatial Randomness ppm(X, ~x) Poisson process with intensity loglinear in $x$ coordinate ppm(X, ~1, Strauss(0.1)) Stationary Strauss process ppm(X, ~x, Strauss(0.1)) Strauss process with conditional intensity loglinear in $x$ } It is also possible to fit models that depend on other covariates.

Manipulating the fitted model:

ll{ plot.ppm Plot the fitted model predict.ppm Compute the spatial trend and conditional intensity of the fitted point process model coef.ppm Extract the fitted model coefficients formula.ppm Extract the trend formula fitted.ppm Compute fitted conditional intensity at quadrature points residuals.ppm Compute point process residuals at quadrature points update.ppm Update the fit vcov.ppm Variance-covariance matrix of estimates rmh.ppm Simulate from fitted model simulate.ppm Simulate from fitted model print.ppm Print basic information about a fitted model summary.ppm Summarise a fitted model effectfun Compute the fitted effect of one covariate logLik.ppm log-likelihood or log-pseudolikelihood anova.ppm Analysis of deviance model.frame.ppm Extract data frame used to fit model model.images Extract spatial data used to fit model model.depends Identify variables in the model as.interact Interpoint interaction component of model fitin Extract fitted interpoint interaction valid.ppm Check the model is a valid point process project.ppm Ensure the model is a valid point process } For model selection, you can also use the generic functions step, drop1 and AIC on fitted point process models. See spatstat.options to control plotting of fitted model. To specify a point process model: The first order ``trend'' of the model is determined by an R language formula. The formula specifies the form of the logarithm of the trend. ll{ ~1 No trend (stationary) ~x Loglinear trend $\lambda(x,y) = \exp(\alpha + \beta x)$ where $x,y$ are Cartesian coordinates ~polynom(x,y,3) Log-cubic polynomial trend ~harmonic(x,y,2) Log-harmonic polynomial trend }

The higher order (``interaction'') components are described by an object of class "interact". Such objects are created by: ll{ Poisson() the Poisson point process AreaInter() Area-interaction process BadGey() multiscale Geyer process DiggleGratton() Diggle-Gratton potential DiggleGatesStibbard() Diggle-Gates-Stibbard potential Fiksel() Fiksel pairwise interaction process Geyer() Geyer's saturation process Hardcore() Hard core process LennardJones() Lennard-Jones potential MultiHard() multitype hard core process MultiStrauss() multitype Strauss process MultiStraussHard() multitype Strauss/hard core process OrdThresh() Ord process, threshold potential Ord() Ord model, user-supplied potential PairPiece() pairwise interaction, piecewise constant Pairwise() pairwise interaction, user-supplied potential SatPiece() Saturated pair model, piecewise constant potential Saturated() Saturated pair model, user-supplied potential Softcore() pairwise interaction, soft core potential Strauss() Strauss process StraussHard() Strauss/hard core point process Triplets() Geyer triplets process } Finer control over model fitting: A quadrature scheme is represented by an object of class "quad". To create a quadrature scheme, typically use quadscheme. ll{ quadscheme default quadrature scheme using rectangular cells or Dirichlet cells pixelquad quadrature scheme based on image pixels quad create an object of class "quad" } To inspect a quadrature scheme: ll{ plot(Q) plot quadrature scheme Q print(Q) print basic information about quadrature scheme Q summary(Q) summary of quadrature scheme Q }

A quadrature scheme consists of data points, dummy points, and weights. To generate dummy points: ll{ default.dummy default pattern of dummy points gridcentres dummy points in a rectangular grid rstrat stratified random dummy pattern spokes radial pattern of dummy points corners dummy points at corners of the window } To compute weights: ll{ gridweights quadrature weights by the grid-counting rule dirichlet.weights quadrature weights are Dirichlet tile areas }

Simulation and goodness-of-fit for fitted models: ll{ rmh.ppm simulate realisations of a fitted model simulate.ppm simulate realisations of a fitted model envelope compute simulation envelopes for a fitted model }

Point process models on a linear network:

An object of class "lpp" represents a pattern of points on a linear network. Point process models can also be fitted to these objects. Currently only Poisson models can be fitted.

ll{ lppm point process model on linear network anova.lppm analysis of deviance for point process model on linear network envelope.lppm simulation envelopes for point process model on linear network predict.lppm model prediction on linear network linim pixel image on linear network plot.linim plot a pixel image on linear network }

V. MODEL FITTING (SPATIAL LOGISTIC REGRESSION)

Logistic regression Pixel-based spatial logistic regression is an alternative technique for analysing spatial point patterns that is widely used in Geographical Information Systems. It is approximately equivalent to fitting a Poisson point process model. In pixel-based logistic regression, the spatial domain is divided into small pixels, the presence or absence of a data point in each pixel is recorded, and logistic regression is used to model the presence/absence indicators as a function of any covariates. Facilities for performing spatial logistic regression are provided in spatstat for comparison purposes. Fitting a spatial logistic regression Spatial logistic regression is performed by the function slrm. Its result is an object of class "slrm". There are many methods for this class, including methods for print, fitted, predict, simulate, anova, coef, logLik, terms, update, formula and vcov. For example, if X is a point pattern (class "ppp"): ll{ command model slrm(X ~ 1) Complete Spatial Randomness slrm(X ~ x) Poisson process with intensity loglinear in $x$ coordinate slrm(X ~ Z) Poisson process with intensity loglinear in covariate Z }

Manipulating a fitted spatial logistic regression ll{ anova.slrm Analysis of deviance coef.slrm Extract fitted coefficients vcov.slrm Variance-covariance matrix of fitted coefficients fitted.slrm Compute fitted probabilities or intensity logLik.slrm Evaluate loglikelihood of fitted model plot.slrm Plot fitted probabilities or intensity predict.slrm Compute predicted probabilities or intensity with new data simulate.slrm Simulate model } There are many other undocumented methods for this class, including methods for print, update, formula and terms. Stepwise model selection is possible using step or stepAIC.

VI. SIMULATION

There are many ways to generate a random point pattern, line segment pattern, pixel image or tessellation in spatstat.

Random point patterns:

ll{ runifpoint generate $n$ independent uniform random points rpoint generate $n$ independent random points rmpoint generate $n$ independent multitype random points rpoispp simulate the (in)homogeneous Poisson point process rmpoispp simulate the (in)homogeneous multitype Poisson point process runifdisc generate $n$ independent uniform random points in disc rstrat stratified random sample of points rsyst systematic random sample (grid) of points rMaternI simulate the Mat'ern Model I inhibition process rMaternII simulate the Mat'ern Model II inhibition process rSSI simulate Simple Sequential Inhibition process rStrauss simulate Strauss process (perfect simulation) rNeymanScott simulate a general Neyman-Scott process rMatClust simulate the Mat'ern Cluster process rThomas simulate the Thomas process rLGCP simulate the log-Gaussian Cox process rGaussPoisson simulate the Gauss-Poisson cluster process rCauchy simulate Neyman-Scott process with Cauchy clusters rVarGamma simulate Neyman-Scott process with Variance Gamma clusters rcell simulate the Baddeley-Silverman cell process runifpointOnLines generate $n$ random points along specified line segments rpoisppOnLines generate Poisson random points along specified line segments } Resampling a point pattern:

ll{ quadratresample block resampling rjitter apply random displacements to points in a pattern rshift random shifting of (subsets of) points rthin random thinning } See also varblock for estimating the variance of a summary statistic by block resampling. Fitted point process models:

If you have fitted a point process model to a point pattern dataset, the fitted model can be simulated.

Cluster process models are fitted by the function kppm yielding an object of class "kppm". To generate one or more simulated realisations of this fitted model, use simulate.kppm.

Gibbs point process models are fitted by the function ppm yielding an object of class "ppm". To generate a simulated realisation of this fitted model, use rmh. To generate one or more simulated realisations of the fitted model, use simulate.ppm.

Other random patterns:

ll{ rlinegrid generate a random array of parallel lines through a window rpoisline simulate the Poisson line process within a window rpoislinetess generate random tessellation using Poisson line process rMosaicSet generate random set by selecting some tiles of a tessellation rMosaicField generate random pixel image by assigning random values in each tile of a tessellation }

Simulation-based inference

ll{ envelope critical envelope for Monte Carlo test of goodness-of-fit qqplot.ppm diagnostic plot for interpoint interaction scan.test spatial scan statistic/test }

VII. TESTS AND DIAGNOSTICS

Classical hypothesis tests: ll{ quadrat.test $\chi^2$ goodness-of-fit test on quadrat counts clarkevans.test Clark and Evans test kstest Kolmogorov-Smirnov goodness-of-fit test bermantest Berman's goodness-of-fit tests envelope critical envelope for Monte Carlo test of goodness-of-fit scan.test spatial scan statistic/test anova.ppm Analysis of Deviance for point process models }

Sensitivity diagnostics:

Classical measures of model sensitivity such as leverage and influence have been adapted to point process models. ll{ leverage.ppm Leverage for point process model influence.ppm Influence for point process model dfbetas.ppm Parameter influence } Residual diagnostics: Residuals for a fitted point process model, and diagnostic plots based on the residuals, were introduced in Baddeley et al (2005) and Baddeley, Rubak and Moller (2011). Type demo(diagnose) for a demonstration of the diagnostics features.

ll{ diagnose.ppm diagnostic plots for spatial trend qqplot.ppm diagnostic Q-Q plot for interpoint interaction residualspaper examples from Baddeley et al (2005) Kcom model compensator of $K$ function Gcom model compensator of $G$ function Kres score residual of $K$ function Gres score residual of $G$ function psst pseudoscore residual of summary function psstA pseudoscore residual of empty space function psstG pseudoscore residual of $G$ function compareFit compare compensators of several fitted models }

Resampling and randomisation procedures

You can build your own tests based on randomisation and resampling using the following capabilities: ll{ quadratresample block resampling rjitter apply random displacements to points in a pattern rshift random shifting of (subsets of) points rthin random thinning }

VIII. DOCUMENTATION

The online manual entries are quite detailed and should be consulted first for information about a particular function. The paper by Baddeley and Turner (2005a) is a brief overview of the package. Baddeley and Turner (2005b) is a more detailed explanation of how to fit point process models to data. Baddeley (2010) is a complete set of notes from a 2-day workshop on the use of spatstat.

Type citation("spatstat") to get these references.

Licence

This library and its documentation are usable under the terms of the "GNU General Public License", a copy of which is distributed with the package.

Acknowledgements

Kasper Klitgaard Berthelsen, Abdollah Jalilian, Marie-Colette van Lieshout, Ege Rubak, Dominic Schuhmacher and Rasmus Waagepetersen made substantial contributions of code. Additional contributions by Ang Qi Wei, Sandro Azaele, Colin Beale, Ricardo Bernhardt, Brad Biggerstaff, Roger Bivand, Florent Bonneu, Julian Burgos, Simon Byers, Ya-Mei Chang, Jianbao Chen, Igor Chernayavsky, Y.C. Chin, Bjarke Christensen, Marcelino de la Cruz, Peter Dalgaard, Peter Diggle, Ian Dryden, Stephen Eglen, Neba Funwi-Gabga, Agnes Gault, Marc Genton, Pavel Grabarnik, C. Graf, Janet Franklin, Ute Hahn, Andrew Hardegen, Mandy Hering, Martin Bogsted Hansen, Martin Hazelton, Juha Heikkinen, Kurt Hornik, Ross Ihaka, Robert John-Chandran, Devin Johnson, Mike Kuhn, Jeff Laake, Robert Lamb, Jonathan Lee, George Leser, Ben Madin, Robert Mark, Jorge Mateu Mahiques, Monia Mahling, Peter McCullagh, Ulf Mehlig, Sebastian Wastl Meyer, Mi Xiangcheng, Jesper Moller, Linda Stougaard Nielsen, Felipe Nunes, Thierry Onkelinx, Evgeni Parilov, Jeff Picka, Sergiy Protsiv, Adrian Raftery, Matt Reiter, Tom Richardson, Brian Ripley, Barry Rowlingson, John Rudge, Aila Sarkka, Katja Schladitz, Bryan Scott, Vadim Shcherbakov, Shen Guochun, Ida-Maria Sintorn, Yong Song, Malte Spiess, Mark Stevenson, Kaspar Stucki, Michael Sumner, P. Surovy, Ben Taylor, Berwin Turlach, Andrew van Burgel, Tobias Verbeke, Alexendre Villers, Hao Wang, H. Wendrock, Jan Wild and Selene Wong.

Details

spatstat is a package for the statistical analysis of spatial data. Currently, it deals mainly with the analysis of patterns of points in the plane. The points may carry auxiliary data (`marks'), and the spatial region in which the points were recorded may have arbitrary shape.

The package supports

  • creation, manipulation and plotting of point patterns
  • exploratory data analysis
  • simulation of point process models
  • parametric model-fitting
  • hypothesis tests and model diagnostics
Apart from two-dimensional point patterns and point processes, spatstat also supports patterns of line segments in two dimensions, point patterns in three dimensions, and multidimensional space-time point patterns. It also supports spatial tessellations and random sets.

The package can fit several types of point process models to a point pattern dataset:

  • Poisson point process models (by Berman-Turner approximate maximum likelihood or by spatial logistic regression)
  • Gibbs/Markov point process models (by Baddeley-Turner approximate maximum pseudolikelihood or Huang-Ogata approximate maximum likelihood)
  • Cox/cluster process models (by Waagepetersen's two-step fitting procedure and minimum contrast)
The models may include spatial trend, dependence on covariates, and complicated interpoint interactions. Models are specified by a formula in the R language, and are fitted using a function analogous to lm and glm. Fitted models can be printed, plotted, predicted, simulated and so on.

References

Baddeley, A. (2010) Analysing spatial point patterns in R. Workshop notes. Version 4.1. CSIRO online technical publication. URL: www.csiro.au/resources/pf16h.html Baddeley, A. and Turner, R. (2005a) Spatstat: an R package for analyzing spatial point patterns. Journal of Statistical Software 12:6, 1--42. URL: www.jstatsoft.org, ISSN: 1548-7660.

Baddeley, A. and Turner, R. (2005b) Modelling spatial point patterns in R. In: A. Baddeley, P. Gregori, J. Mateu, R. Stoica, and D. Stoyan, editors, Case Studies in Spatial Point Pattern Modelling, Lecture Notes in Statistics number 185. Pages 23--74. Springer-Verlag, New York, 2006. ISBN: 0-387-28311-0.

Baddeley, A., Turner, R., Moller, J. and Hazelton, M. (2005) Residual analysis for spatial point processes. Journal of the Royal Statistical Society, Series B 67, 617--666.

Baddeley, A., Rubak, E. and Moller, J. (2011) Score, pseudo-score and residual diagnostics for spatial point process models. To appear in Statistical Science.

Diggle, P.J. (2003) Statistical analysis of spatial point patterns, Second edition. Arnold.

Gelfand, A.E., Diggle, P.J., Fuentes, M. and Guttorp, P., editors (2010) Handbook of Spatial Statistics. CRC Press.

Huang, F. and Ogata, Y. (1999) Improvements of the maximum pseudo-likelihood estimators in various spatial statistical models. Journal of Computational and Graphical Statistics 8, 510--530.

Waagepetersen, R. An estimating function approach to inference for inhomogeneous Neyman-Scott processes. Biometrics 63 (2007) 252--258.