Message-ID: <CCE622E7.11DB9%jchandler@emiconsulting.com>
Date: 2012-12-06T18:29:46Z
From: Jess Chandler
Subject: presence-only data analysis
In-Reply-To: <CALSKosB-XkXtLVOxoS-ZeZ3HXU7Ujga1F3iTWipWBgmQnOO=xA@mail.gmail.com>
SH,
What you are looking at sounds like a hazard rate analysis (but with only
location and a dummy variable). You can define a set of neighbor counties
and presence for those counties (or set of counties) as another variable.
You will then have
Outbreak risk = a(presence) + b(location-maybe) + c(neighbor)
Which model in R that you use will depend upon whether you have time
series data.
Jess
On 12/6/12 10:21 AM, "SH" <emptican at gmail.com> wrote:
>Dear List:
>
>I have googled to try to understand analyzing presence-only data and
>predicting (?) species outbreaks. Although I was able to find a few
>papers, it is somewhat hard for me to figure out adapting my data.
>Or, I may be looking at a wrong place. Any suggestions/comments will
>be appreciated.
>
>I have a data set of GPS coordinates (long, lat) and presence-only
>data (i.e., '1') based on counties. I would like to estimate
>likelihood of outbreak index. For example, we know there is an
>observation in (say) county A (or A, B, and C neighboring counties)
>and I would like to know the likelihood (e.g., probabilities) of other
>surrounding or neighboring counties (e.g., D, E, F, etc.). Could you
>give me any suggestion on analyzing this type of data? I DON'T have
>any covariates.
>
>Thank you in advance,
>
>SH
>
>_______________________________________________
>R-sig-Geo mailing list
>R-sig-Geo at r-project.org
>https://stat.ethz.ch/mailman/listinfo/r-sig-geo