• Very short introduction (these slides).

• Live tutorial with examples of analysis.

• Questions and discussion.

• 25 years development by mainly Adrian Baddeley with co-author Rolf Turner from the beginning and more recently I joined the team.
• 100,000+ lines of R code, 30,000+ lines of C code, 2,500+ exported objects and 1000+ documentation files.
• Recently split into sub-packages spatstat.xxxx, which the now almost empty package spatstat then Depends on.

• Strong focus on backwards compatibility.
• Follows base R design and uses base R graphics.
• Extensive usage of the S3 object system.
• Has its own S3 objects for data etc.
• Detailed documentation.
• Use ?spatstat to get an overview and find hidden gems.

• You can use spatstat to describe/summarise any given point set with things like

  • Pairwise distances, nearest neighbour distances, empty space distances, Dirichlet/Voronoi tessellations, …

• However, spatstat really focuses on statistical inference for phenomena that generate random locations (point processes).

library(spatstat)
set.seed(42) # Reproducibility
Xpois <- rpoispp(100, nsim = 3)
plot(Xpois, main = "")

Xhc <- rHardcore(beta = 100, R = .05, nsim = 3)
plot(Xhc, main = "")

Xthomas <- rThomas(kappa = 5, mu = 20, scale = .1, nsim = 3)
plot(Xthomas, main = "")

lambda <- function(x,y){200*(x^2+y^2)}
Xinhom <- rpoispp(lambda, nsim = 3)
plot(Xinhom, main = "")

• Intensity is a first moment property.

• Interaction is a higher moment property (inter-point correlation).

• They are confounded and without further assumptions it is impossible to separate them in general.

• Often a approach like in time series is used:

  • First, model the mean (trend, seasonality).
  • Second, model the interaction after accounting for the mean model.

• Explore ?spatstat which includes lists of commonly (and less commonly) used functions.

• Get the book. Unfortunately we don’t have a license to share an online version as many authors have nowadays. Maybe this will change with a second edition. There are three free sample chapters at https://book.spatstat.org/

• Ask questions on stackoverflow under the spatstat tag

• Report bugs or make feature requests on GitHub. (If possible find the right sub package repo to put the issue under.)