gs.test.semipar: Semiparametric two-sample testing

In neurodata/graphstats: Graph Statistics

Description

This is a simple implementation of the semiparametric two-sample testing problem of given X_1, X_2, …, X_n i.i.d. F and Y_1, Y_2, …, Y_m i.i.d. G, test the null hypothesis of F = G against the alternative hypothesis of F \not = G.

Usage

gs.test.semipar(g1, g2, k = NULL, nsamples = 2 * max(gorder(g1),
  gorder(g2)), test = "R", verbose = FALSE)

Arguments

g1	input graph, as an igraph object. See graph for details.
g2	input graph, as an igraph object. See graph for details.
k	dimension of the latent position that graphs are embeded into. Defaults to NULL which selects the optimal dimensionality using gs.dim.select.
nsamples	Number of bootstrap samples when performing hypothesis tesing. Defaults to 2*n where n is the number of vertices, max(gorder(g1), gorder(g2)).
test	the type of test statistic to use. Defaults to 'R'. Supported options are: 'R'estimate only a rotation between the latent positions of g1 and g2 when computing the test statistic.
verbose	logical indicating whether to print output to console. Defaults to FALSE.

Value

a list containing the following:

T.obs	The observed test statistic for whether g1 and g2 are sampled from the same distribution. If T.obs is near 0, this indicates that g1 and g2 likely come from the same distribution.
T.null	A length nsamples vector indicating the test statistic under the bootstrapped samples for whether g1 and g2 are sampled from the same distribution.
p.value	the p-value associated with the semiparametric two-sample test; the fraction of times T.alt < T.obs.

Author(s)

Eric Bridgeford [e-mail]

References

Athreya, A., Fishkind, D. E., Levin, K., Lyzinski, V., Park, Y., Qin, Y., Sussman, D. L., Tang, M., Vogelstein, J. T., Priebe, C. E. Statistical inference on random dot product graphs: a survey

neurodata/graphstats documentation built on May 14, 2019, 5:19 p.m.