lsmi_union {snowboot} | R Documentation |
Obtain one big LSMI – with max(n.seeds)
seeds and n.wave
waves around each – and subsample seeds to create smaller LSMIs (with less
seeds and/or waves). The function is primarily used in cross-validation.
lsmi_union(net, n.seeds, n.wave, seeds = NULL)
net |
a network object that is a list containing:
The network object can be simulated by |
n.seeds |
an integer vector of numbers of seeds for snowball sampling
(cf. a single integer |
n.wave |
an integer defining the number of waves (order of the neighborhood)
to be recorded around the seed in the LSMI. For example, |
seeds |
a vector of numeric IDs of pre-specified seeds. If specified, LSMIs are constructed around each such seed. |
Note that the produced LSMIs are slightly different from those described by Gel et al. (2017). The current R implementation produces smaller LSMIs by subsetting the seeds, not by new sampling of seeds from the network and growing completely new LSMIs, as it was done by Gel et al. (2017). See the details in Figure 3 by Chen et al. (2018)
A list with two elements:
lsmi_big |
LSMI with |
sequence_seeds |
A list of length equal to |
Chen Y, Gel YR, Lyubchich V, Nezafati K (2018).
“Snowboot: bootstrap methods for network inference.”
The R Journal, 10(2), 95–113.
doi: 10.32614/RJ-2018-056.
Gel YR, Lyubchich V, Ramirez Ramirez LL (2017).
“Bootstrap quantification of estimation uncertainties in network degree distributions.”
Scientific Reports, 7, 5807.
doi: 10.1038/s41598-017-05885-x.
sample_about_one_seed
, lsmi
, lsmi_cv
net <- artificial_networks[[1]]
a <- lsmi_union(net, n.seeds = c(5, 10, 15), n.wave = 2)