[R-sig-eco] Using partykit::ctree with observations that are spatially autocorrelated

Hello list members, I'm planning to model species presence / absence data (N = 523) that was collected over a geographic area. Samples come from preferential sites (sea level > 1200 m, obligatory presence of permanent waterbodies, etc). I want to infere on environmental factors determining the occ

[R-sig-eco] Using partykit::ctree with observations that are spatially autocorrelated

thanks manuel, i'm afraid unmarked can not deal with spatial autocorrelation - see this post: http://groups.google.com/group/unmarked/browse_thread/thread/f45bd5ab12aa127d/26446409f3e2a140?lnk=gst&q=spatial%23#26446409f3e2a140 best, kay 2012/3/20 Manuel Spínola > You need to look in the "unma

[R-sig-eco] Using partykit::ctree with observations that are spatially autocorrelated, edit

Hello list members, I'm planning to model species presence / absence data (N = 523) that was collected over a geographic area. Samples come from preferential sites (sea level > 1200 m, obligatory presence of permanent waterbodies, etc). I want to infere on environmental factors determining the occ

Re: [R-sig-eco] Using partykit::ctree with observations that are spatially autocorrelated

That is true, but it can deal with detectability, and I am afraid ctree cannot. Best, Manuel 2012/3/21 Kay Cichini > thanks manuel, > > i'm afraid unmarked can not deal with spatial autocorrelation - > see this post: > http://groups.google.com/group/unmarked/browse_thread/thread/f45bd5ab12aa12

Re: [R-sig-eco] Using partykit::ctree with observations that are spatially autocorrelated

manuel, the sites where sampled repeatedly, such that the detection probability is sufficiently high. some occurrences might be overseen, but as i am not interested in the abundances or occurrence-rate per se i can live with missing some. another problem of regression like models is that i would

Re: [R-sig-eco] Using partykit::ctree with observations that are spatially autocorrelated

many thanks bálint - this one helped me a lot! in the case of a binomial outcome, would you then get the residuals by taking the difference of observed and predicted average occurrence-rate per node? thanks, kay ps: i'll foward this to the list.. 2012/3/21 Bálint Czúcz > Hi, Kay, > you can co

Re: [R-sig-eco] Using partykit::ctree with observations that are spatially autocorrelated

What you say makes sense to me Kay. Best, Manuel 2012/3/21 Kay Cichini > manuel, > > the sites where sampled repeatedly, such that the detection probability is > sufficiently high. some occurrences might be overseen, but as i am not > interested in the abundances or occurrence-rate per se i ca