# problem with enveloped test in spatstat

5 messages
Open this post in threaded view
|

## problem with enveloped test in spatstat

 Can any *spatstat* user explain to me why the two p-values obtained below for an envelope test against CSR are so different? > data(swedishpines) > d<-swedishpines > plot(d) > mad.test(d,Kest,nsim=999,verbose=F)         Maximum absolute deviation test of CSR         Monte Carlo test based on 999 simulations         Summary function: K(r)         Reference function: theoretical         Alternative: two.sided         Interval of distance values: [0, 24] units (one unit = 0.1 metres)         Test statistic: Maximum absolute deviation         Deviation = observed minus theoretical data:  d mad = 150.69, rank = 216, p-value = *0.216* > mad.test(d,*Lest*,nsim=999,verbose=F)         Maximum absolute deviation test of CSR         Monte Carlo test based on 999 simulations         Summary function: L(r)         Reference function: theoretical         Alternative: two.sided         Interval of distance values: [0, 24] units (one unit = 0.1 metres)         Test statistic: Maximum absolute deviation         Deviation = observed minus theoretical data:  d mad = 2.9921, rank = 9, p-value = *0.009* *!!!* These data are dispersed relative to CSR: >kl<-envelope(d,Kest,nsim=999,correction="border") >plot(kl) Dave Unwin         [[alternative HTML version deleted]] _______________________________________________ R-sig-Geo mailing list [hidden email] https://stat.ethz.ch/mailman/listinfo/r-sig-geo
Open this post in threaded view
|

## Re: problem with enveloped test in spatstat

 Hi David, This is very clearly explained in page 393 of Baddeley et al. 2015. Basically, the MAD test is affected by transformations of the summary function, with the L-function providing more powerful tests because of its stabilization of the variance. If you are interested in point pattern analysis I would recommend you to get a copy of this nice book. Cheers, Marcelino Adrian Baddeley, Ege Rubak, Rolf Turner (2015). Spatial Point Patterns: Methodology and Applications with R. London: Chapman and    Hall/CRC Press. http://www.crcpress.com/Spatial-Point-Patterns-Methodology-and-Applications-with-R/Baddeley-Rubak-Turner/9781482210200/  El 21/04/2018 a las 18:05, David Unwin escribió: > Can any *spatstat* user explain to me why the two p-values obtained below > for an envelope test against CSR are so different? > > > >> data(swedishpines) >> d<-swedishpines >> plot(d) >> mad.test(d,Kest,nsim=999,verbose=F) > > >          Maximum absolute deviation test of CSR > >          Monte Carlo test based on 999 simulations > >          Summary function: K(r) > >          Reference function: theoretical > >          Alternative: two.sided > >          Interval of distance values: [0, 24] units (one unit = 0.1 metres) > >          Test statistic: Maximum absolute deviation > >          Deviation = observed minus theoretical > > > > data:  d > > mad = 150.69, rank = 216, p-value = *0.216* > > > >> mad.test(d,*Lest*,nsim=999,verbose=F) > > >          Maximum absolute deviation test of CSR > >          Monte Carlo test based on 999 simulations > >          Summary function: L(r) > >          Reference function: theoretical > >          Alternative: two.sided > >          Interval of distance values: [0, 24] units (one unit = 0.1 metres) > >          Test statistic: Maximum absolute deviation > >          Deviation = observed minus theoretical > > > > data:  d > > mad = 2.9921, rank = 9, p-value = *0.009* > > > > *!!!* > > > > These data are dispersed relative to CSR: > >> kl<-envelope(d,Kest,nsim=999,correction="border") >> plot(kl) > > Dave Unwin > > [[alternative HTML version deleted]] > > _______________________________________________ > R-sig-Geo mailing list > [hidden email] > https://stat.ethz.ch/mailman/listinfo/r-sig-geo> . > -- Marcelino de la Cruz Rot Depto. de Biología y Geología Física y Química Inorgánica Universidad Rey Juan Carlos Móstoles España _______________________________________________ R-sig-Geo mailing list [hidden email] https://stat.ethz.ch/mailman/listinfo/r-sig-geo
Open this post in threaded view
|

## Re: problem with enveloped test in spatstat

 Many thanks.  It is a superb book and amazing value for money.  My only excuse is that I have not as yet got to page 392.  Plotting the functions also makes it a lot clearer. Dave On 21 April 2018 at 22:50, Marcelino de la Cruz Rot < [hidden email]> wrote: > Hi David, > > This is very clearly explained in page 393 of Baddeley et al. 2015. > > Basically, the MAD test is affected by transformations of the summary > function, with the L-function providing more powerful tests because of its > stabilization of the variance. > > If you are interested in point pattern analysis I would recommend you to > get a copy of this nice book. > > > Cheers, > > Marcelino > > > Adrian Baddeley, Ege Rubak, Rolf Turner (2015). Spatial Point Patterns: > Methodology and Applications with R. London: Chapman and >   Hall/CRC Press. > http://www.crcpress.com/Spatial-Point-Patterns-Methodology-> and-Applications-with-R/Baddeley-Rubak-Turner/9781482210200/ > > > > > > > El 21/04/2018 a las 18:05, David Unwin escribió: > >> Can any *spatstat* user explain to me why the two p-values obtained below >> for an envelope test against CSR are so different? >> >> >> >> data(swedishpines) >>> d<-swedishpines >>> plot(d) >>> mad.test(d,Kest,nsim=999,verbose=F) >>> >> >> >>          Maximum absolute deviation test of CSR >> >>          Monte Carlo test based on 999 simulations >> >>          Summary function: K(r) >> >>          Reference function: theoretical >> >>          Alternative: two.sided >> >>          Interval of distance values: [0, 24] units (one unit = 0.1 >> metres) >> >>          Test statistic: Maximum absolute deviation >> >>          Deviation = observed minus theoretical >> >> >> >> data:  d >> >> mad = 150.69, rank = 216, p-value = *0.216* >> >> >> >> mad.test(d,*Lest*,nsim=999,verbose=F) >>> >> >> >>          Maximum absolute deviation test of CSR >> >>          Monte Carlo test based on 999 simulations >> >>          Summary function: L(r) >> >>          Reference function: theoretical >> >>          Alternative: two.sided >> >>          Interval of distance values: [0, 24] units (one unit = 0.1 >> metres) >> >>          Test statistic: Maximum absolute deviation >> >>          Deviation = observed minus theoretical >> >> >> >> data:  d >> >> mad = 2.9921, rank = 9, p-value = *0.009* >> >> >> >> *!!!* >> >> >> >> These data are dispersed relative to CSR: >> >> kl<-envelope(d,Kest,nsim=999,correction="border") >>> plot(kl) >>> >> >> Dave Unwin >> >>         [[alternative HTML version deleted]] >> >> _______________________________________________ >> R-sig-Geo mailing list >> [hidden email] >> https://stat.ethz.ch/mailman/listinfo/r-sig-geo>> . >> >> > -- > Marcelino de la Cruz Rot > Depto. de Biología y Geología > Física y Química Inorgánica > Universidad Rey Juan Carlos > Móstoles España > > -- David J. Unwin Professor Emeritus in Geography Birkbeck, University of London Phone  +44(0)1604 686526 Mobile: +44(0)7840 297239 (text preferred) SKYPE: david.unwin99         [[alternative HTML version deleted]] _______________________________________________ R-sig-Geo mailing list [hidden email] https://stat.ethz.ch/mailman/listinfo/r-sig-geo