56 reputation
3
bio website
location
age
visits member for 1 year, 7 months
seen Mar 5 at 16:14

Jul
8
awarded  Popular Question
Feb
7
comment Anyone have recommendations on data collection system for bird point counts?
I will check it out. Thank you!
Feb
6
asked Anyone have recommendations on data collection system for bird point counts?
Jan
30
awarded  Supporter
Jan
30
comment What is an appropriate statistic to measure spatial autocorrelation of points with binary values?
Yes, this is related to sampling biases, and the suggestion to compare distributions of the covariates at sample points and globally seems sensible. But, I am also concerned about whether these biases effectively lead to some level of pseudoreplication, and how this affects the model stability or accuracy. I have never seen any suggestion in the literature that RF removes the assumption of random, independent samples, which is an assumption even for standard non-parametric statistics.
Jan
30
comment What is an appropriate statistic to measure spatial autocorrelation of points with binary values?
I should clarify... What I am most interested in is determining how problematic the clustering in my training data is likely to be in terms of generating a RF model. I have approximately 20 presence points and 70 absence points, but some of these points are very closely spaced on the landscape (as little as 200 m of another presence or absence point). In the past I have used some filtering to eliminate closely spaced points, but wonder if that is necessary since the nearby points tend to represent 1 presence and 1 absence.
Jan
29
comment What is an appropriate statistic to measure spatial autocorrelation of points with binary values?
I'm constructing a species distribution model, and have both presence and reliable absence locations for the species. There is clear clustering of points on the landscape, but perhaps little spatial autocorrelation, since presence and absence points occur in close proximity. In other words, our samples are spatially biased, with points occurring within hundreds of meters of other points, but if spatial autocorrelation is not significant, I'm wondering how much the clustering matters in terms of modeling, inferences, etc. I'm using Random Forest for modeling.
Jan
29
comment What is an appropriate statistic to measure spatial autocorrelation of points with binary values?
Thanks, Jeffrey. Joint counts are clearly the way to go for binary data, but I saw a suggestion (can't recall where, now) that joint counts were only appropriate for area (not point) data. It wasn't apparent to me why you couldn't create the weights matrix using a distance threshold, and use a joint count, but I couldn't find examples of this in some cursory searches. Is there a reference you can provide for this type of use?
Jan
29
asked What is an appropriate statistic to measure spatial autocorrelation of points with binary values?
Dec
21
comment How can I merge the attribute tables from feature classes into a single table, easily?
I started down this road. ArcMap bombs out on me when building the ModelBuilder model, let alone running it. The datasets are so large and numerous that the 32-bit memory limitations kick in.
Dec
21
awarded  Student
Dec
20
asked How can I merge the attribute tables from feature classes into a single table, easily?