[R-sig-Geo] Temporal marked point process with time-varying covariates

[Mayeul KAUFFMANN]

Thanks a lot!
The authors of the RLadyBug package are also about to release functions to
estimate a temporal marked point process with time-varying covariates and both
time and space correlation (probably in a few weeks).
They informed me by private e-mail but stated they were not opposed to my
putting this piece of information here.
Best regards,
Mayeul

_____________________________________________________
Dr. Mayeul KAUFFMANN, Conflict Specialist
European Commission, Joint Research Centre (JRC)
Institute for the Protection and Security of the Citizen (IPSC)
Global Security and Crisis Management - ISFEREA
Via E. Fermi 2749 - I-21027 Ispra (VA), ITALY
Phone: (+39) 033278 5071
http://isferea.jrc.ec.europa.eu/Staff/Pages/Kauffmann-Mayeul.aspx

(Office: building 48c, 1st floor, room 123. TP: 483)

-----Original Message-----
From: Adrian.Baddeley at csiro.au [mailto:Adrian.Baddeley at csiro.au] 
Sent: Thursday, August 26, 2010 10:18 AM
To: mayeul.kauffmann at jrc.ec.europa.eu; r-sig-geo at stat.math.ethz.ch
Subject: RE: Temporal marked point process with time-varying covariates

Mayeul KAUFFMANN wrote:
 
> I am trying to model a temporal marked point process with time-varying
> covariates and I am looking for the most appropriate function among several
> ones. [ ...] I had a look at the following packages 
>    spatstat
>    splancs
>    PtProcess
> spatstat seems to have the correct object to handle my dependant variables
(the
> ppx class: 2D space + time) but if I'm correct the ppm() model fitting
function
> cannot handle this (it only works with ppp). Am I missing something? I saw at
> http://www.spatstat.org/  that this branch is in development. Any news /
> schedule on that?

Yes, the class 'ppx' in spatstat will support space-time point pattern data with
any number of space and time dimensions. 

Currently the model-fitting function will only handle two-dimensional point
patterns (class 'ppp'). However ppm will soon be able to handle ppx objects. We
have code, but it is not ready for release yet. 

Due to some bad experiences in the past, I am reluctant to release spatstat code
that involves original research until the research papers have been published. 

Just to clarify something: 'spatstat' is **not** committed to a particular
definition of the conditional intensity. If the points are in time or
space-time, where time is one-dimensional, then the natural definition of the
conditional intensity is one which looks at the 'past'. However if the points
are in m-dimensional Euclidean space, then the most appropriate definition of
the conditional intensity is something different (usually the Papangelou
conditional intensity). In spatstat, the type of conditional intensity is
determined by the 'interaction' argument to ppm (or actually by
'interaction$family'), and thus can be different from model to model. Currently
ppm deals with two-dimensional point patterns and the interactions mostly use
the Papangelou conditional intensity, but the package design does not make any
such assumptions. [We also have code for fitting models that use the directed
conditional intensity in two-dimensional time, and this will be released
shortly.]

Adrian Baddeley