hypervolume_overlap_confidence: Confidence intervals for overlap statistics

In hypervolume: High Dimensional Geometry, Set Operations, Projection, and Inference Using Kernel Density Estimation, Support Vector Machines, and Convex Hulls

Confidence intervals for overlap statistics

Description

Generates confidence intervals of four different overlap statistics. In order to find the confidence interval for the overlap statistics of two hypervolumes, use hypervolume_resample twice to generate bootstraps. The function takes in paths to two sets of bootstrapped hypervolumes and gets overlap statistics for each possible pair. Confidence interval is calculated by taking a quantile of generated overlap statistics.

Usage

hypervolume_overlap_confidence(path1, path2, CI = .95, cores = 1)

Arguments

path1	A path to a directory of bootstrapped hypervolumes
path2	A path to a directory of bootstrapped hypervolumes
CI	Desired confidence interval proportion
cores	Number of logical cores to use while generating overlap statistics. If parallel backend already registered to doParallel, function will use that backend and ignore the argument in cores.

Details

The four overlap statistics are Sorensen, Jaccard, frac_unique_1, frac_unique_2. See hypervolume_overlap_statistics

Each hypervolume from path1 is overlapped with each hypervolume from path2 using hypervolume_set. The four overlap statistics are calculated for each overlap.

Value

jaccard	Confidence interval for jaccard similarity score
sorensen	Confidence interval for sorensen similarity score
frac_unique_1	Confidence interval for fraction of first hypervolume that is unique
frac_unique_2	Confidence interval for fraction of second hypervolume that is unique
distribution	a matrix of overlap statistics used to generate the confidence intervals

See Also

hypervolume_resample

Examples

## Not run: 
# Let us overlap two hypervolumes generated from multivariate nomral 
# distributions with different means and same covariance matrices.
sample1 = rmvnorm(150, mean = c(0, 0))
sample2 = rmvnorm(150, mean = c(0.5, 0.5))

hv1 = hypervolume(sample1)
hv2 = hypervolume(sample2)

# generates confidence intervals from quantiles of 20*20 overlaps
path1 = hypervolume_resample("mean_0_0", hv1, n = 20)
path2 = hypervolume_resample("mean_0.5_0.5", hv2, n = 20)

result = hypervolume_overlap_confidence(path1, path2)
# confidence index of Sorensen coefficient
print(result["sorensen"])


## End(Not run)

hypervolume documentation built on Sept. 14, 2023, 5:08 p.m.