[R-sig-Geo] plotting Ripley fonctions

Adrian Baddeley adrian at maths.uwa.edu.au
Tue Oct 14 07:39:42 CEST 2008

Lassana Toure wrote:

 > I need some help for plotting L(r) and Lcross so that I can have my
 > confidence interval(CI) parallel to X axis.
 > The plot I have obtain so far the graphic is not horizontal, then it
 > is difficult to interpret strong the aggregation is or the attraction.

This is a question about the package 'spatstat'.

Presumably you want to plot L(r) - r against r, so that a horizontal 
line corresponds to a completely random pattern; and you want the 
envelopes to be horizontal as well.

If X is your point pattern, then type

     E <- envelope(X, Lest, global=TRUE, nsim=39)
     plot(E, . - r ~ r)

This is for L(r) = sqrt(K(r)/pi) where K(r) is Ripley's K-function. To 
do the same thing for the cross-type L function of a marked point 
pattern, e.g. for the 'amacrine' data:
      V <- envelope(amacrine, Lcross, i="on", j="off",
                       global=TRUE, nsim=39)
     plot(V, . - r ~ r)

For further information, see the course notes on spatstat which are 
available at <www.spatstat.org>.

Please note that simulation envelopes are NOT confidence intervals for 
the true value of L(r). They are critical bands for a test of the 
hypothesis of a completely random pattern.

Adrian Baddeley
co-author of spatstat

More information about the R-sig-Geo mailing list