overlap: Overlap between the prior and posterior distribution

In birdring: Methods to Analyse Ring Re-Encounter Data

Overlap between the prior and posterior distribution

Description

Gives the overlap of two distributions (such as a prior and a posterior distribution) based on one sample of simulated values from each distribution

Usage

overlap(posterior, prior, from = 0, to = 1, nsim = 1e+05, edge.of.parameter.space=FALSE)

Arguments

posterior	A numeric vector, a sample of simulated random values from the posterior distribution
prior	A numeric vector, a sample of simulated random values from the prior distribution
from	Lower limit of the parameter space over which the posterior and prior distributions are compared.
to	Upper limit of the parameter space over which the posterior and prior distributions are compared.
nsim	Number of simulated values used for the Monte Carlo simulation to measure the overlap.
edge.of.parameter.space	logical value; Two different methods are implemented to calculate the overlap. First (edge.of.parameter.space=FALSE), smoothers are fitted to the histograms of the simulated values from the posterior and prior distributions, and the overlap is calculated based on this smoothed density functions. This has the advantage to be more exact when the number of simulated values from the posterior distribution is small. However, it can be unreliable when the mean of the posterior distribution is close to the edge of the parameter space. In such cases (edge.of.parameter.space=TRUE), it is more reliable to calculate the overlap directly from histograms of the simulated values from the posterior and prior distributions. See also details.

Details

If edge.of.parameter.space=FALSE, the function first uses the function density to obtain density functions of the prior and posterior distributions and then the overlap is measured by a Monte Carlo simulation. If edge.of.parameter.space=TRUE, two histrograms of the simulated values from the posterior and prior distributions are drawn with 999 classes and breaks 1000 equally spaced values between from and to. The overlap is then calculated directly from these histograms.

Value

a numeric value which is an approximation of the proportion of the overlap of the posterior with the prior distribution.

Author(s)

Fraenzi Korner-Nievergelt

References

Gimenez, O., S. P. Brooks, et al. (2009). Weak identiability in models for mark-recapture-recovery data. Modelling Demographic Processes in Marked Populations. Series: Environmental and Ecological Statistics. D. L. Thomson, E. G. Cooch and M. J. Conroy.

See Also

density

Examples

prior <- rbeta(2000, 1,1)
posterior <- rbeta(2000, 14, 35)
overlap(posterior, prior)

birdring documentation built on Oct. 7, 2023, 5:07 p.m.