hypervolume (version 3.1.4)

hypervolume_funnel: Hypervolumes at different sample sizes

Description

This function takes in hypervolumes bootstrapped at different sample sizes applies a function to each hypervolume. The output of the function can either be a plot of nonparametric confidence intervals or a table of the mean and quantiles.

Usage

hypervolume_funnel(input_path, 
                    title = NULL, 
                    func = get_volume, 
                    CI = .95, 
                    as_table = FALSE)

Value

ggplot object, or dataframe object

Arguments

input_path

output of resample with method = "bootstrap seq"; path to a sequence of different sample sized bootstraps

title

title of output plot, ignore if outputting as table

func

a function that takes a single parameter which is a hypervolume and returns a numerical value.

CI

Confidence interval is taken by using the the (1-CI)/2 and (1+CI)/2 quantile

as_table

If TRUE, returns a table with columns upper quantile, mean, lower quantile

Details

This function is used to evaluate the behavior of hypervolumes at different sample sizes and determine bias. Statistics such as volume are affected by sample size especially when the hypervolumes are constructed with method = "gaussian" since the bandwidth estimate is dependent on sample size.

Examples

if (FALSE) {
# 3000 data point hypervolume
data(quercus)
hv_quercus = hypervolume(quercus[,c(2,3)])

# the seq argument is equivalent to a length 30 vector {10, 139, ... , 3649, 3779}
# 6hr sequential runtime
quercus_bootstrap_seq <- resample('quercus_bootstrap_seq', 
                                    hv_quercus, 
                                    method = 'bootstrap seq', 
                                    points_per_resample = "sample_size", 
                                    seq = floor(seq(10, 3779, length.out = 30)),
                                    cores = 20)

# Compatible with ggplot syntax when used with as_table = FALSE
hypervolume_funnel(quercus_bootstrap_seq, 
                    title = 'Resampled volumes of Quercus', 
                    func = get_volume) + 
  geom_line(aes(y = get_volume(hv_quercus))) +
  ylab("Volume")
}