Hello,
I am currently evaluating different ways to analyse a specific multiple
species monitoring system. One idea is to identify the important covariates
with random forest. However, there are some points I have problems to get my
head around.
1. I have observations from a GRID with 25 systematic plots, as well as 5
random plots within the GRID. To analyze observations of a single species
with rf, can I pool those 30 plots together, or is this invalidly mixing of
random and systematic plots?
2. The systematic plots are 100 m apart from each other. I have distance
data for each observation (DISTANCE sampling is another tool used). If I
truncate the data at 50 m, I get truly independent data for each plot (apart
from the 5 random plots, that is). Is there an advantage doing it this way,
or is it more plausible to use untruncated data and have more observations?
3. As I said, I have multiple species. Is there a difference between running
rf analysis on data of a single species only (presence/absence) versus
pooling all species into one analysis and separating the results?
4. For birds, I have aural and visual detections. This has a big influence
on detectability, so I keep analysis in DISTANCE apart. I assume that I can
pool them together for rf if I use aural/visual as a covariate, right?
Any comments are greatly appreciated.
Best regards,
Dirk Nemitz
--
---------------------------------------
Dirk Nemitz
Zeppelinstr. 11a
37083 Göttingen
Germany
Tel: +49 (0)551 492 32 51
Mobil: +49 (0)175 709 31 92
---------------------------------------
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
