gambin (version 2.4.0)

summary.gambin: Summarising the results of a gambin model fit

Description

S3 method for class 'gambin'. summary.gambin creates summary statistics for objects of class 'gambin'.The summary method generates more useful information (e.g. confidence intervals) for the user than the standard model fitting function. Another S3 method (print.summary.gambin; not documented) is used to print the output.

Usage

# S3 method for gambin
summary(object, confint = FALSE, n = 50, ...)

Arguments

object

A gambin model fit object from fit_abundances

confint

A logical argument specifying whether confidence intervals should be calculated (via bootstrapping) for the parameters of gambin models with more than 1 component (confidence intervals for 1 component gambin models are calculated automatically)

n

The number of bootstrap samples to use in generating the confidence intervals (for multimodal gambin models)

…

Further arguments to pass

Value

A list of class 'summary.gambin' with nine elements, containing useful information about the model fit.

Details

For the one-component gambin model the confidence interval for the alpha parameter is calculated automatically using an analytical solution.

For gambin models with more than one component no analytical solution for deriving the confidence intervals is known. Instead, a bootstrapping procedure can be used (using the confint and n arguments) to generate confidence intervals around the alpha and max octave parameters. However, the process can be time-consuming, particularly for gambin models with more than two components. Thus, the default is that confidence intervals are not automatically calculated for gambin models with more than one component (i.e. confint == FALSE).

In addition, it should be noted that in certain case the confidence intervals around the alpha parameters in multi-component gambin models can be quite wide. This is due to changes in the max octaves of the component distributions in the bootstrapped samples. It can be useful to make a plot (e.g. a dependency boxplot) of the n alpha values against the max octave values.

Examples

# NOT RUN {
data(moths)
fit = fit_abundances(moths)
summary(fit)
# multimodal gambin models with confidence intervals
biMod <- fit_abundances(moths, no_of_components = 2)
summary(biMod, confint = TRUE, n = 5) #large n takes a long time to run
# }