Practical Model Building for Quantitative Population Ethology
with Event-Driven Competing Risks
by
Ideas have been presented elsewhere on a framework for quantitative
population ethology using event-driven competing risks. In today's
climate, it is not enough to have an idea--one must have an
implementation of the idea. To that end, we present the beginnings
of such a system built on a public domain, statistical system called
R. We use spline-based graphical tools to craft the mean value
functions of the competing risk structure, based either on data or
on prior belief. Spatial location of individuals is developed on a
hexagonal grid at the resolution of the model, with a triangular
coordinate system to reduce computation of distances. The components
can be quickly developed and tested in R, and later translated to C
for more efficient coding of the most highly used loops. We also
discuss some of the design requirements that affect the development
of an efficient competing risk model of large problems.
Click to get
manuscript. The figures are not quite
complete yet [Figure 1 was redone; Figure 4 is missing; Figure 9 is
out of place], although this should be rectified presently.