ppgam
generalised additive point process models
about
This R package allows fitting of point process models in which the parameters have generalised additive model form.
citation
Youngman, B. D. and T. Economou (2017). Generalised additive point process models for natural hazard occurrence. Environmetrics 28 (4), e2444. DOI: 10.1002/env.2444.
download
Here's the latest source code. (This is fairly basic code that will need a lot of work before it can be submitted to CRAN. This may get done... eventually.)
example 1: spatial point process model
Here's an R script with an example of fitting a point process to locations of storm tracks occurring in the North Atlantic. The point process rate varies over space using a using a thin-plate regression spline. The code produces a map of the estimated (log) rate, which is shown below.
example 2: spatio-temporal point process model
This example extends example 1 by allowing spatial-temporal variation such that the spatial point process rate varies by time of year. This is achieved by additive thin-plate and cyclic cubic regression splines. The latter ensures continuity from 31st December to 1st January. The code produces a map of the point process rate for different times of the year, which is shown in the animation below.
example 3: spatio-temporal-NAO point process model
This example extends example 2 by allowing spatial-temporal variation in the point process rate to also depend on the corresponding state of the North Atlantic Oscillation (NAO). We use an interaction between spatial and NAO variation, which is achieved by a tensor product between thin-plate and cubic regression splines. The code produces a map of the point process rate for different NAO values for 1st January, as in the animation below.
